David Holz, founder of AI art generator Midjourney, on the future of imaging In 2008, David Holz co-founded a hardware peripheral firm called Leap Motion. He ran it until last year when he left to create Midjourey. Midjourney in its present form is a social network for creating AI-generated art from a text prompt – type a word or phrase at the input prompt and you'll receive an interesting or perhaps …
Always a problem
Obviously there are some legal restrictions - in UK not even remotely photorealistic stuff e.g. cartoons, can be regarded as kiddie pron & would still be a massive legal offence (though some artists, e.g. Belmer, have made a career / cash from creating things your average member of the public would serve jail time for) *
But beyond the obviously illegal, always a slippery slope as lots of people will use this to create art (you may lack painting skills but if you can craft an appropriate phrase AI may do something you would have liked to paint)
I like a variety of artists, but being UK based may as well reference "recent" British artists where some of their works have caused controversy, I imagine the moderators would not like images that resemble some pieces by Bacon, Freud, Jake & Dinos etc.
I can see all these art AI companies going the chocolates box cover art style only.
* Remember seeing some Belmer** stuff in the UK years ago (I did not enjoy it, but looking at challenging art sometimes does make you feel queasy), not sure if it would still be allowed with current rules as the old art / pron argument never goes away & sometimes curators go the path of least legal resistance.
** I really do not advise looking this up online as miles beyond NSFW, if you don't know his work, be grateful
However appreciating art is only half the story. The reason we create is to express ourselves and our emotions. I code and do system admin by day but I do photography in my spare time as a way to express my creativity, to find something that makes me feel some emotional connection to the world around me. Much as I love a superb, effcient system creating perfect art from the command line might all well and good but plugging in values, letting some AI algorithm determine what's good and not, well it's not the same as pouring your heart and soul into something you created and feel and emotional attachment to.
As a media studies professor who has never had the technical skill to create the images that I can conjure and describe in my mind, this tool is liberating!
I understand your point about pouring your heart and soul into the artwork, and it is weird that artwork can be generated so quickly with this tool. On the other hand, I have been fiddling and tweaking images AA I learn about the tool, and this process feels much like my creative art making process when working with other media. It is just that my output looks better.
But for me this tool is expanding the range of possibilities both professionally and creatively. For people who make art for a living, I can understand the anxiety about ai art.
I don't know my emotional reaction to a splodge of paint on canvas until I've splodged it. One can't control splats and drips, clouds of ink in water, or marbellings of oil on water. The physical world is computing an output from physical input. A paint brush held between my toes makes the painting mine. Tying a brush to a tree in a breeze... the work is mine? Tying a brush to a stoat and letting it run across a whitewashed floor... my work or the stoat's?
Physicists, philosophers and artists may be happy with the ambiguity. The IP lawyers (and the stoat) less so.
The surrealist painter H R Geiger often used airbrushes in his work to convey the sheen of shiny objects, a technique common in pre-CGI commercial illustration, especially of shiny cars. Used with skill it can give a near photorealistic appearance to a painting.
He reports once having had difficulty with some country's customs, having to explain to the officials that his "obscene photographs" were actually paintings, which were legal. Geiger's response was "Yeah right, they're photographs - where the hell do you think I took them?"
There's sure to be soon a ML thingy that can create and light a 3D environment such that a player's point of view approximates these images. A never-ending dreamscape. Combined with the flood of 3D data about our real lived-in environments that will soon come from digital twin models and AR sensors. Alice Through The Looking Glass In Basingstoke, Fear And Loathing In Bournemouth. Nightmare fuel.
So in the 2 trials I had access to I attempted to get an illustration of "Boris Johnson waving a British flag leaning out of a Delorean Time Machine" for a post I wanted to write regarding him leaving office.
Not a single one was a Delorean, just a great box, flags were often in the background, and the cartoon likeness of Boris was usually melded with an abstract grey car body.
The nature of Midjourney running on Discord means they have user limits, meaning they're going to start removing accounts, plus they have no other payment options yet.
I'm holding hope that DallE will be better with regards to image generation, but their t&c won't allow me to specify Boris Johnson, so I'd have to add "male blonde prime minister" or something to get an abstraction for my post - which I never ended up writing but at least it became a good benchmark into the tech.
@Ben 56
Try Craiyon
Tried that just now with prompt "Boris Johnson waving a British flag leaning out of a Delorean" & got OK results
Maybe also look at Wombo - results tend to be poor for complex queries like yours though but last time I tried it gave Johnson images on a simple query
.. both a bit more basic however in training data and output quality
It does mention this, but I have seen a few artists observing that by taking their work and incorporating it into the source model for the AI it's a kind of diffuse copyright-theft. Without their art, the AI would be worthless, but the outcome of the AI is likely to be losing them work and offering no benefit to them.
Also what happens as we have more AI generated art online? Do we end up with AI just recycling AI produced-images on a slowly entropic path, as we're starting to see with text generation online?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
