I watched this video back in May 2022 and even though the human artist takes more time, and Dall-e has a couple of interesting compositions, it’s really clear that the computer is not a threat to his ability to produce to the spec convincingly. I’m not sure if before this video I had give much thought to AI generated art.
I was vaguely aware of things like snapchat filters, people putting their faces into an ap and getting back their likeness as a cartoon or anime or Disney character, although I was disinterested in giving those aps my camera data, and I’ve used several avatar generators where you assemble the parts like a digital stickers version of a fancy looking potato-head doll, or like making your wii avatar. Novel. But all of this is ultimately just like a bizarre as seen on tv kitchen gadget, meaning it’s of limited usefulness. At best, I imagined the latest “AI” art, with the guidance and the over-painting of an artist, could be a tool for editorial illustration or book covers.
At the time, it wasn’t really a big topic in the world of artists overall from my perspective. Maybe AI Art had come up in some of my art newsletters which tend to be a little ahead of the digital art waves, such as the Blender Guru guy with the doughnut modeling lessons.
Someday, I always tell myself, I’ll try to make that doughnut properly again… Some day.
In August 2022, a friend sent me a link to mid journey, and I joined the discord. I marveled like everyone else as I watched the prompts go by, some yielding splendor or horror or cuteness or quirkiness. I didn’t know what it was all about. I made a few free prompts which came back with underwhelming results, probably in part because I wasn’t leveraging some of the more detailed prompt controls.
It’s always derivative by definition, but it’s sometimes derivative in a surprising way. So, like someone pulling on a slot machine arm, the prompters are sort of mining the mash up possibilities and hoping for some winners among the dross. The process is also quite a time suck, even though the number of images on return is many orders greater than anyone can produce by hand, you’re still sat pulling the slot machine lever trying to tune to what you’re actually looking to get.
After the transfixing spell of watching the scroll of midjourney discord results wore off for me, I didn’t really look at it again for months.
Not long after, another friend sent me a link to a less impressive but unlimited free AI image generator called craiyon, and by reigning in my expectations I managed to get some cute low res things. This too wore off quickly. We joked around about the disappointing and horrifying results of trying to get the craiyon AI to make figures. In one set, he had asked it to draw a particular person in a pencil sketch style, and looking at the disfigured work emulated in pencil style, I could tell that the computer was not correctly resolving proportions, most likely because it couldn’t accurately map the face— yet. As an artist, I know that struggle. It runs deep into my past and still plagues me, almost as much as hands and gestures. “The computer looks like it’s in about 7th grade,” I said. “Tell it to follow it’s dreams.” I was joking when I typed it, but I went away knowing there was a lot of truth in it. The computer would keep trying and it would develop better accuracy. It likely already had in other iterations of the technology.
My experience by that point in November led me to decide that the currently publicly available versions of this so called AI Art (a misnomer I’ll discuss as I carry on my thoughts) is not really that scary, and could be very useful in meeting a market that artists can’t keep up with and earn a fair wage at— A lower priced one of making little avatars and thumbnails and mood board and inspiration art.
I had done, and still occasionally will do, small paid digital painting and drawing commissions for friends’ character art. These paintings have taken me anywhere from 6 to 100 hours each. The higher end of my pricing is typically $75. You do the math on my hourly rate. With out something to cut down that time, this is not a viable business model or even any significant side hustle. I’m thankful for the extra cash, and often use it to buy myself art supplies and electronic goodies, making my hobby pay for itself. But it’s only hobby money. I put aside doing these commissions for the most part, with the intention to learn how to use free image resources and photobash what would amount to an under-drawing so I could cut some time out by doing a less intensive level of paint over process. My hope was that if I learn photobashing, I could maybe half my input time. But realistically I know that it would sort of just shift it from painting to resource gathering and image stitching. Hint: this is pretty close to what AI is basically doing at this time. Although results may vary, for better or for worse, in comparison with a human making the source image and stitching choices himself.
When I thought about the underwhelming results I was getting back from some of my midjourney prompts or the mangled or not quite natural ways some parts of other people’s prompts were coming back, I realized, there’s a new market for hobbiest painters/ photoshoppers or entry level folks. People could potentially hire artists to retouch their generated images, or commissioners could send an artist AI generated images as a mood board, so it takes out some of the part where the artist is trying to read the commissioner’s mind. Both of those approaches seemed reasonable at my level and sort of opens another kind of entry level or hobby level market (read: in no way a living wage). And artists who don’t want the work or find it too demeaning can take a pass on it.
Hence, in August ’22, I still wasn’t really fussed by midjourney or craiyon. Prompt art had a lot of limitations and it potentially filled a market (that market being just anyone with an idea and pocket change) for which there are not enough artists to work at for the small potatoes those interested parties could pay. But it’s not developed enough for serious commercial use anyway. The quality can potentially be shiny, but the capacity for filling a brief or working with consistency is going to require a larger set of tools and a new kind of practice in the pipeline of art process. It’s possible, but not currently there.
And it was clear that the computer is just basically doing some more sophisticated version of applying filters. It’s taking from its dataset, analyzing patterns and then weighting different parts of it’s resources to blend them according to the keywords in the prompt. This is just a form of machine learning, since it gets prompts and then builds from feedback of selected responses and hones it’s model. It can certainly become more sophisticated, and sometimes because of the mashup nature of the process, it turns up an image that appears to be surprising (hard to say it’s something that’s never been done, since we aren’t seeing the images it’s sourcing from along side it). But it’s not the same process of learning and thinking that a human does. It’s not really what I would call “Intelligence”.
A human mind has context for their creations- they have a culture and an upbringing, a family and a history and an education, an experience of embodiment, and a spiritual life and dreams and friends. Intelligence involves a kind of deep consciousness that isn’t just signified by the skill of pattern making and generative barf from a dataset. The machine doesn’t consider meaning and life or have feelings or ideas. It’s just rehashing based on some algorithms. It seems smart, but it’s just a fancy version of google-image-search-meets-photo-filters.
But no matter what it is, it is a disruption. For good or for evil?
By the end of 2022, my feeling was, and in general still is, it’s like the advent of Photography or 3-D art or the internet.
The camera didn’t make artists obsolete, but rather became a huge gift to artists. At first I can see how it might have looked like a challenge to turn of the century portrait artists and landscape artists wholesale. But in fact, the market for taking an image recording was much much larger than all of the painters in all of human history with 100% of their time and effort could ever meet. The average person could now have pictures of their kid’s birthday party. It was no longer limited to the courts of kings. And the resource of photographs became an incredible source for painters to use to work from as well. Painting continued and a new field of art in photography was opened up. It went through a second watershed with the improvement of digital photography. No longer was the average home photographer precious about how many frames was left on the film roll, and with immediate displaying results, re-shoots and creativity could blossom too. An entirely new category of art thrives now and not only that, but it improved the work of painters who could benefit from easy sources of reference for training and self improvement. Some painters no longer felt the need to be representational, and it allowed painting to be more than just a hireling capturing the likeness of his patrons. Painting almost demanded the painter reach for expression in order to set itself apart from the ubiquitous photography. Whole new movements in style resulted.
In the same way, I know I can’t possibly paint all of the original character ideas my friends have for their amateur fiction or personal fantasy worlds. It could consume all of my time, even presuming money was no object, and I still couldn’t meet that demand in my life time. Watching my friends replace their avatars and face claims for characters and fill their moodboards is for me a sense of joy and relief. I’m happy for them. And I’m also reminded that I can develop my skill to say what it is I want to say on my own projects and that the human made art will likely stand out in a sea of generated stuff. At least if I’m doing it right and reaching for what I genuinely want to express.
It’s possible this tool is going to be another multiplier in the artistic process for those who chose to use it. It’s possible that just like googling reference images, we could use this tool essentially as a google mash up for mood boards and parts to source for our own painting. I was content with the idea that it would probably be another adjustment with more benefits than drawbacks.
3-D art turned the effects industry upside down. People who got onboard with the new tech and developed it became the new leaders in fx work. People who wanted to remain with animation, model building, physical fx, and muppets continued to improve their own categories of style and while the boom chased the new shiny balls in computer generated art, the traditional methods held up too. 2-D digital tools grew easier and more accessible and animators largely moved to digital pipelines. Filming cameras improved for digital imaging, and then everything could be pipelined in multiple stages in computer formats. Often we see a mix of strategies to manage different parts of a production. We have more tools and more ways to get a vision across to the audience. Those productions that limit themselves to traditional methods now stand out as an artistic vision for doing so.
The internet was a huge boon to creative artists, and massively disruptive. Two ways I see it having affected me as an artist the most are in education and distribution. I went to artschool and slogged my way through and the product (my skills) was not that amazing when I graduated. In the mid 2010’s, A few years after I graduated from a SUNY visual arts program, I wanted to get back into drawing and started self-teaching by following internet tutorials. The results were amazing. In a short period of time (under a year) I improved at a level the university couldn’t have even offered to an undergrad (there are other aspects of personal development the university focused on that were not skills based- intellectual development or humanities exposure, essentially. much of this is also easy to access online now, although it would be important to have help curating material or assistance to create your own curriculum and learn to research. I digress). Online, I was able to customize what I was looking to learn rather than choose from the limited offerings of a university and to pay professionals to tutor me and to get feedback from professional art communities offering their help. The sheer amount of searchable visual resources available to draw from (and to literally draw from) were far beyond anything I was able to collect from clipping magazines and taking my own photos. I found myself jealous of my kids as I shared my love of drawing with them. They were able to develop their art skills basically 20 years younger than I was able to. (no resentment- more like amazement) and I know this is because of the sheer quality and quantity of professional grade, accessible teaching and high quality index searchable visual aid available. For free.
If I would have known what was on the horizon, I might have kept my money spent on university and spent my time differently and just upped my skills when the advent came about. But I have a piece of paper issued by SUNY. so. yay. I guess if I hadn’t earned it, I wouldn’t have known that I wasn’t missing out. It’s an expensive lesson that I get to share.
Then there is the opportunity online to publish at the push of a button. Anyone can post their work, make and sell anything in an online shop, advertise their physical sales, make digital assets, get commissions etc. Artists who can leverage the internet as a publishing tool and promotional tool have more freedom and success available to them than most artists a generation ago. This isn’t to say artists are all rolling in the dough or that every artist can leverage it. I just mean that it made it possible for more artists to thrive in a new way that never existed to them before.
So yeah, that was all that generally went into thinking about the first whiff of this so called revolution of AI art. Essentially, I figured people should still draw and paint and some people were gonna get new mileage out of leveraging the tool just like past ones. Put a pin in this initial conclusion. It’s still pretty close to what I think. But some new stuff came up I should cover too. More for a part two!
Nikki Dreamer Illustration 
Leave a comment