Reasons to make art with AI

Gepubliceerd op 4 februari 2023 om 15:00

First reason: I want to create something

I can never stop making up stories, but it's a lot of work to also write down those stories, or draw them on paper, or convert them into a working game. The AI that are now emerging have the potential to accelerate this process. I started with a lot of questions: what is possible, can it generate my own style, can it make all the details I want, how controllable is it? More importantly: what is my own style? And: which AI do I want to use?

When push comes to shove, you go Googling and that's how I ended up at NightCafe Creator, where they offer a number of AI under one roof (so you can experiment at leisure), and they have a pleasant user interface (so you don't have to program yourself or give cryptic control commands via Discord). As with many platforms, the first trials are free, but you can't avoid paying if you really want to get started. After all, nothing in life is free, not even computer time. It turns out to be very addictive to conjure up the most beautiful creations at the touch of a button. My conclusion quickly was: the AI can draw better than me. Now I have to try to get my intended main character and my intended Dutch(!) locations drawn. Here I quickly ran into the first bumps…

On bike after school in Delft

Second reason: Search for my own style

My first goal is to be able to illustrate a story - a self-written story with self-drawn or self-generated illustrations. I think up my stories quite visually, so I know roughly how I would like to draw: detailed, with color but not overpowering (like Hergé with Tintin?), a slightly sinister atmosphere but suitable for 12-year-olds (like Arthur Rackham?), both everyday and fantasy themed (like Rien Poortvliet?), science fiction and steampunk…. Like many aspiring writers, I have a tendency to want to cram too much into one story. The good news is: the AI is familiar with all genres and a lot of artists. The question is: how well can they be mixed, so that an original 'Harmanna style' emerges?


UFO grabs piece of Pillars of creation

Then there emerge a lot of buttons to turn. What you want to depict is just the beginning. Then comes the "how". Materials: chalk, watercolor, oil paint, airbrush, or maybe…? Style: art deco, art nouveau, rococo, storybook illustration, or maybe…? Inspired by which artists, or studios, or platforms? Hyper-detailed, or Ukiyo-e? Photorealistic or animation? Limited color palette or rainbow colors? Use terms like epic, cinematic, aether punk, or eldritch if you want it to be big and compelling. Also use technical terms such as ambient occlusion and volumetric lighting for extra atmospheric lighting. Use traditional painting terms, photography terms, cinematic terms, or terms from the computer graphics and game industry.

This is just the tip of the iceberg. And of course - everything has to be in English, but typos are okay, the algorithms often guess the intended word.

Third reason: Immerse yourself in art

Abstract art in bright colors

While searching for the terms you want to use for your own style, you will automatically immerse yourself in the various art movements. (Unless you know in advance that you only want the modern Japanese 'studio Ghibli' style, or a 'Pixar' style, because if you just want to copy, you'll be done quickly.)

And not just movements, but also materials, techniques, old masters, new masters. The latest AI are trained with everything that can be found online, so if you want to know more about it, you can just Google it. With a few words you can express your ideas in Pop Art or Renaissance style, in cyberpunk neon colors or black and white pen drawing, like Disney or like Rembrandt - as far as the AI can guess. Do you want to know how the AI interprets Kandinsky's work? Then just enter "by Kandinsky" as a prompt and see what happens.

Woops - that way you will quickly burn a lot of computer time and therefore money.

Fortunately, many colleagues have their prompts open, so you can also see which terms they use for their best results. You will also discover terms that are not among the defaults that NightCafe recommends, new artists, new tricks for atmospheric effects. Gradually you will probably also discover that there is more than one style that appeals to you. You might want to use one style to illustrate your book, two other styles to create wall art, and a fourth to make Christmas greeting cards. No one will stop you!

Other people's reasons: self-expression

I've reflected enough on my own self-expression now, but I see people on the platform with other reasons for using AI to create art, beyond the fun of creating itself.


First of all these are the professionals: indepen-dent artists and artists who make concept art for the film and game industry. They can use AI to quickly generate a series of ideas, test them, and then build on them.

Kawaii baby Cthulu

Self knowledge

In addition, there are those who seem to seek self knowledge or self-actualization through self-expression: I see (probably) younger people generating image after image of the ideal woman/goddess/personification-in-female-form (and sometimes male), and applaud the most beautiful results. These are probably young women (it's not always easy to determine with all those self-chosen avatars and usernames) looking for their ideal, imagining where they themselves want to be in ten years - perhaps also with whom they want to be in ten years. With them I often miss images of old (and wise) men and women, but I imagine that those will come naturally.

Horror genre


A third category are people who seem to be engaged in therapy: they generate either horror, or dreamscapes with and without cute bunnies and unicorns. A few say it explicitly in their profile description (one is processing his PTSD), but I suspect there are more given the large number of horror enthusiasts on the one hand and chibi enthusiasts on the other. (Or these are just two schools of thought I can't quite grasp.)

Not safe for work

A fourth category are lovers of nudity, but they can't go to just any AI. A large portion of the publicly available text-to-image generators prohibit anything that might be offensive; a few ask the user to indicate which works are not suitable for a general public by means of a 'Not Safe For Work' indication. Criminal images such as child pornography are also not allowed in the latter, and are countered as much as possible by filtering prompts before the images are generated.

Much more is possible

What I make myself and show on this website are by no means the only things that are possible. Some artists are way more creative than me, or just have more extreme taste - who knows. Where the AI ensures that the talent for crafts (drawing, painting, photography) becomes less important, the talent for having original ideas only becomes more important. Here, in my opinion, lies the source of real, original art.

Dire wolf in moonlight

Reactie plaatsen


Er zijn geen reacties geplaatst.

Maak jouw eigen website met JouwWeb