Over the past couple of months, I've seen a lot of artists on Twitter experimenting with AI and art. This is a very fast-developing thing right now. There have been multiple online software programs launched that do it in a similar fashion: you type in a few words or a phrase and the AI returns two or three images. One AI image was given an award in the Digital Image category in a recent juried exhibition, which caused a stink among many artists. So what's it all about?
The primary AI generators, at least the ones that I've seen images from most often, are DALL-E, Stable Diffusion, and MidJourney. There are many many others. Results can be stunning. I see a lot that look like sci-fi landscapes or urbanscapes: impossible architecture, cities floating in the sky, barren horizons, Star Wars-like figures, or a single figure silhouetted against an ominous sky. Others seem like Old Masters paintings that have been bumped up to 11. Many are extremely realistic, others less so. Here are a few examples:
When I compared AI images to oil paintings, I thought that they had some really interesting aspects. They made good use of composition. See how the padre is highlighted by the arch, or how the single figure in Returnlessness is offset to one side and balanced by the light pole. They can be highly detailed, far more than the typical oil painting is taken (Returnlessness again), or they can suggest details without actually portraying them (Busy). All the AI images I've seen have been representational - that is, they show recognizable things: people, buildings, light poles, and so on. None of it has been abstract, although they certainly make use of good abstract principles.
I tried out a couple of programs, but got really awful results. And I didn't like the process. It felt like pulling the lever on a slot machine over and over, only instead of a money, you might get a nice image. There was nothing of "me" in it.
It seemed to me (and still seems, come to think of it) that AI is good for a "Wow" effect, like great eye candy. But what keeps bugging me is that it's machine made. It can't have soul, but it can simulate it, sometimes well.
Having said that, I have found one artist who is doing really good work with AI. Francien Krieg, a Dutch artist, is the exception that proves the rule. She has been doing a series of oil paintings using herself and old women as her subjects. Earlier this year, she started playing with AI and she seems to have figured out how to get it to do images that are very similar to her oil paintings. Here's an example:
Francien is the only one I have seen who is able to give her AI images character, depth, and most importantly, heart. You really should check out her website and compare her physical paintings to her AI work.
As for me, I will stick with physical artworks. But I like seeing what other artists are creating with AI. Many of the NFTs that I've been collecting are AI-generated. In fact. all of the ones in this post are NFTs that I've bought. So, yeah, I think there's a place for AI art. Just not in my studio.
No comments:
Post a Comment