As someone who once dreamed of being an illustrator, all of this text-to-image generator stuff makes me heart bleed and spurt in panic, terror and devastating sadness. As someone whose kids are both artistic and love to draw and let images and expression just ooze out of their tiny hands onto paper in colours, shapes and quite wacky compositions, my heart bleeds even more. In fact, it’s squeezed and pulped like an orange on a Sunday morning. It seems the “bros,” whoever and wherever they may be, have killed, that is belittled, commodified and rendered magic-less, art. Maybe.
The machine sucks up not just pixels but decades of learning, experience, experimenting and inordinate pain of the artist. It doesn’t care. They scrape and ingest pixels and nothing more, caring not for the decades of practice and centuries of learning that artists have produced. They have an insatiable appetite for more and more pixels. They don’t care about someone taking years to develop even a rudimentary observational skill to understand light, shade and tone. The companies who made it and tossed it into the world without a second thought don’t care either. This is what makes me angry.
The cat is out of the bag for sure, and there’s no putting back the ability for Bob in Accounting to post his masterpieces on Midjourney of “Bob from Accounting on the beach with Pamela Anderson circa 1995 totally in love and totally partying in the style of Frank Frazetta.” By the way, if you were wondering no generator can handle this particular prompt that I’ve seen so far.
There is another possible way to reframe this and have a way for artists and in particular illustrators remain at least slightly relevant to humanity. This is to think of these tools as replacing human work and expertise, but in the grand service of the artist. That is to think of these AI tools as being the minions of an artist’s studio.
The Studio and the Artist
Rembrandt was reported to have had a huge studio that resembled a life size 1:1 painting factory where his underlings and understudies would do a lot of the heavy lifting on his works such as “The Nights Watch” (1642).
That means they would do what at that point in his career, things he would think of as drudgery, let’s say painting in backgrounds like clouds and trees which are hard and boring to do. The studio system was a training system where the young initiates did this sort of stuff. They might even if they were good, work on a human figure or two, or lay the base tones. They might work from his sketches while he waltzed in a couple times a week and put the final details and gloss on a black hatted Amsterdam resident. Maybe the studio assistants would do other thing that were hard, like hands and hair. Things Rembrandt might not have time for or couldn’t be bothered with. The idea might be, that machines do all this work now, and we should start thinking about them as true assistants and not replacements. For instance tools like SceneryAI can do things that sort of point of this.
As disconcerting and exciting things are happening in terms of the creation of digital imagery, we need to remember there are a number of things we as artists and designers can do about it.
We can and should try to take back at least a smidgen of control with tools that might keep your style and legacy of work being engorged by a training model. This should only be part of it. We as artists and designers have been very bad at this bit, but there are thankfully some tools-minded and humanity-minded developers out there who are able and have begun to setup some infrastructure to address this. I would suggest as a next step that they add a button to auto-sue whoever may have scraped and ingested your work as an arists.
Artists and designers need to get at least a handle on things, but they can’t fight the tides, and so we also need to take part in the design of these tools. This means designing the interfaces and interactions to create the imagery. This is not only reimagining better ways of doing out-painting, but doing things like creating the masterpieces with help to make the figures with the lighting you can adjust, and colour grading the scenes and even generating the clouds and background to match. What would be in the palettes for these sorts of interactions? Would they even be like Photoshop? Would one even have menus? With the new interaction methods that we could imagine, we can also reimagine what needs to happen to support a new, potentially old, way of working. If the artist now has at their disposal the workers to accomplish bits of work for them, do they then have brushes for detailed painting and then brushes to show where the AI should be doing something else? What does the tweaking, sliders and controls for mixed human and machine creation look like? This is what I want to see, and something I should probably get around to design. It is alas, New Years Eve and I am suddenly half in the bag. Let’s take this over.