[generic] To A.I. or not to A.I., that is the question

#1
So whether we like it or not, artificial intelligence and contextual processing by software algorithms is here to stay.

Obviously, we know that in Late State Capitalist Societies, the monolithic corporate leviathan's like Amazon, Google, Meta, or Microsoft will latch unto anything that allows them to reduce paying actual humans for work.

Anyway, small Indie-Press publishers too have chosen A.I. art for its cost savings, but now some of them face some serious pushback from the community, sadly, very negative expressions of that push back on social media.

I am thinking that maybe like the music industry with those "Parental Advisory" labels, maybe we get some voluntary "Made with A.I." advisory labels.

Let us consider two legally viable use cases, those that are not direct cloning/plagiarizing existing human art.

(A) Art generated by an A.I. software system with prompts but none that actually name drop a human artist, just a genre and keywords of the desired output art.

(B) Art first generated by A.I. for inspiration to render scenes never seen before but still extrapolated from human art works and then process by a human artist.


So if the artists source used by the A.I. for its training are an unknown, that is their names were not explicitly fed into the A.I., then should we just consider that open attribution, in so much as the fact it is the human viewers who then try to ascribe an artist's name and style to the A.I. work, which may or may not be wrong.

label-ai-art-open-attribution.png or the other option for processed art label-ai-art-process-by-humans.png

For example below is an explanation given to a curious would-be-purchase of a book with some A.I. art.
And I'll need to give a bit of context and explanation to answer in a meaningful way. So, yes, the commercial version of Midjourney 4 was used by our in-house pro artist to concept, draft, and initially generate the baseline imagery components used in LAIR. They were then heavily manipulated, assembled, corrected, and processed, with substantial additional manual overpainting, in post-production by the same artist to achieve the final result - no small amount of work requiring no small amount of skill. There were no artist names used in prompts, and all illos were paid for at the market rate.
When it comes to copying existing art, this has happened between human artists too.

I am sure we heard of Roy Lichtenstein and his iconic art of exploding military jets.

"WHAAM!" painted in 1963, was my favorite in childhood and I considered Roy Lichtenstein an icon.

https://www.tate.org.uk/art/artworks/lichtenstein-whaam-t00897

The work’s composition is taken from a panel drawn by Irv Novick which appeared in issue number 89 of All-American Men of War,
published by DC Comics in February 1962.

1684872019541.png


But investigations by journalists years ago and more recently revealed his work was plagiarized from other artists' works.

https://www.theguardian.com/artandd...ations-of-plagiarism-against-roy-lichtenstein

1684871677634.png
 
#2
Speaking of inspiration, here is a Kickstarter I am following, Niobe, with some cover art homage to the iconic piece by Frank Frazetta's "Conan the Barbarian"

https://www.frazettagirls.com/products/conan-the-barbarian-print

So, I think the biggest problem with A.I. generated art is the vast mishmash of influences that erodes the attribution to particular artists.
Maybe we can use A.I. to track down A.I. influences and pass A.I. art through a filter to seek attribution like we do for plagiarism in universities.

1684873509629.png
 

torus

Rune Priest
#3
Yes my argument against AI art is not that it uses prior work, which as you say is what all artists do at some level. It's that it is invariably mediocre at best, and more often just awful and in a creepy way.

And the problem is that many of the people enthusiastic about it seem to be the kinds of people who never really placed much value on good art by humans either. Some were already unable to tell the difference between a good illustration or painting and a bad one. Already some very prolific RPG artists are quite patently unable to draw. RPG companies are small businesses and it's a lottery whether the art direction is being handled by someone with any actual talent in this area.

Therefore all AI does in this context is make it even easier for untalented artists, and people with no aesthetic sense whatsoever, to produce terrible illustrations.

Call me snobby perhaps. You might say what does it even mean if some art, AI or otherwise, is considered 'good' or not? Surely if I like it that's all that matters. Art has always been plagued by this problem of subjectivity. And sure, that's fine. You can say the same about writing; and popularity seems to have little to do with talent in that sphere too. I'm confident that R.A. Salvatore could be replaced with an AI with absolutely no bother to anyone besides himself, and probably a marked improvement in quality.

Well, I certainly accept people's right to like bad art - there are certain Sci-Fi artists I quite like myself that fall into this category. But I would hate if artists of real talent were essentially squeezed out by the automation of mediocrity, and that's what I envisage happening with AI art.
 
#4
DriveThruRPG is okay with A.I. generated art.

Here is a step-by-step guide for using MidJourney in your next Anime project.
One thing, I noticed about A.I. is that the girls tend to be very pretty. Is that bias?

1684874088049.png


And a public notice from the author about A.I. processed art.
1684874013661.png
 
#6
Yes my argument against AI art is not that it uses prior work, which as you say is what all artists do at some level. It's that it is invariably mediocre at best, and more often just awful and in a creepy way.

And the problem is that many of the people enthusiastic about it seem to be the kinds of people who never really placed much value on good art by humans either. Some were already unable to tell the difference between a good illustration or painting and a bad one. Already some very prolific RPG artists are quite patently unable to draw. RPG companies are small businesses and it's a lottery whether the art direction is being handled by someone with any actual talent in this area.

Therefore all AI does in this context is make it even easier for untalented artists, and people with no aesthetic sense whatsoever, to produce terrible illustrations.

Call me snobby perhaps. You might say what does it even mean if some art, AI or otherwise, is considered 'good' or not? Surely if I like it that's all that matters. Art has always been plagued by this problem of subjectivity. And sure, that's fine. You can say the same about writing; and popularity seems to have little to do with talent in that sphere too. I'm confident that R.A. Salvatore could be replaced with an AI with absolutely no bother to anyone besides himself, and probably a marked improvement in quality.

Well, I certainly accept people's right to like bad art - there are certain Sci-Fi artists I quite like myself that fall into this category. But I would hate if artists of real talent were essentially squeezed out by the automation of mediocrity, and that's what I envisage happening with AI art.
I agree with you in that human art has value, and if A.I. makes it impossible to make a living as a human painter, the world and or gaming spaces will be that much poorer.

For example, human artists create a distinctive style that enables their fans to follow their work.

I for one got to know Wayne Reynolds ever since he did the covers for my childhood RPG Keith Baker's Eberron for D&D 3.5e.

And I noticed the style in 4th Edition cover art too, and simultaneously also on Pathfinder book covers (lucky Wayne to have Wizards and Paizo both paying him for covers at the same time).

And below is the latest Kickstarter by Wayne Reynolds with folks like me paying for just sketches and happy to help a human.

I also backed Larry Elmore's Kickstarter, especially for his Black and White drawings. His art graced my Dragonlance novels.

1684874688516.png
 
#7
An interesting proposition this "id.art" to have a copy of your art for cross-referencing, authentication, and backup. But I wonder what their stance would be on A.I. art and plagiarism?

1684957392223.png
 
#8
Midjourney A.I. is strange in that sometimes it gets closer what a good human artist would draw with regards to abdominal musculature for anthropomorphic creatures, but other times the A.I. fails woefully, and all generated from the very same text-to-image prompts.

ai-art-tiger-man-iteration-1.png
 

Attachments

#9
With my zoologist hat on... tigers have neither tusks nor horns, so its definitely better at abdominal muscles than at feline dentition. :)

AI art in RPGs... I'm sad if it is depriving people of a living.

But given that I and the Snipping Tool spent considerable portions of my life swiping definitely-not-my-copyright photos of actors, images from comics, pretty landscape photos, etc off the interwebs to use as game reference pics, I'd be a hypocrite if I criticised other GMs or player for using AI to do more or less the same thing.
 
#10
With my zoologist hat on... tigers have neither tusks nor horns, so its definitely better at abdominal muscles than at feline dentition. :)

AI art in RPGs... I'm sad if it is depriving people of a living.

But given that I and the Snipping Tool spent considerable portions of my life swiping definitely-not-my-copyright photos of actors, images from comics, pretty landscape photos, etc off the interwebs to use as game reference pics, I'd be a hypocrite if I criticised other GMs or player for using AI to do more or less the same thing.
Using images for profit changes things (I too have used pictures from the internet for probably two decades in the play by post games I run) but I'd say using images snipped from the internet for gaming is fine (free advertising!).

My own take is that AI art can be a threat (artists rise up with banners and refuse to work because computers are nicking their livelihoods) or a tool (the artist takes the base picture which can be quite 'wow! but... how many fingers?' and fixes it up using their skill which is what the Lair of The Leopard Empresses artist did). I lean towards the latter but then I don't make my living drawing and painting (I'd starve to death based on my ability...)

From my point of view I see Pros:

Fast! No more waiting weeks/months for art from humans who have lives, problems, issues
Imaginative - some of the stuff is 'how did they think of that'???
You can refine/hone to your exact spec without costing a fortune in artist time and ending up with something 'that will do' because you are skint
For $30/month Midjourney (which looks to be about the best) is still cheaper than human art for your indie rpg so you don't have to use stock art or 'drawn with your wrong hand?' own art

Cons:

Competes with human artists, deprives of business/livelihood
Potential sales lost as some consumers won't touch it
How many fingers? Do people really have three arms? WTF am I looking at? - Some of the stuff is weird/wrong/broken
Free AI isn't great. It can be ok, but needs work or use just a bit and ditch the rest.

It's early days. We're going to see so many RPGs come out with AI stuff that people will have to decide if they want to go with the flow and buy it or stick to their guns and put their money towards the corporate stuff from WOTC, Paizo etc who have taken a stance (certainly Paizo has anyway) not to use AI art. If the big companies start using it too... crikey I can imagine various forums exploding.
 
#11
In general, I have tried to stay away from the AI debate, as I have no real skin in the game, but you know, Friday morning and all...

1. People have always 'been inspired' by other compositions. This is not new, nor a defense for piracy.
2. The side narrative being woven into the AI debate - that somehow clipping or right click+save-ing art online for private use in games is somehow denying an artist income - is nonsense and a red herring.

but now onto point 3, but first a history lesson. Storytime!

Some of you might well know that I used to be involved in the UK indie rpg scene. Do you ever really stop? Anyway, back in the day when I produced my own game, I often said that it cost '£7.95 and a bottle of vodka' - a one-month sub to clipart.com and a tip to my mate for doing the character illustrations. And I had to force that onto him! A little bit of photoshop magic, a smattering of Indesign, and I had a game. Nobody has ever complained about the state of the books - but they are what they are, a relic of a former age. Black and white, perfect bound, softcover books with line art and not a whole lot of it.

Now, cut forward to today. Kickstarter has changed the game. As RPGers we have become used to full colour, big ass hardback books once again. And we can have them because Kickstarter gives people who can be arsed with it the chance to have their games funded to the tune of £000s. Of course, you then have to deal with art direction, professional layout, negotiating with printers in Estonia and such, but the returns can be lucrative. And all power to the people who do that?

But what about the hobbyist designer? How can they compete in this old-is-new world of Big Bold Books? Do they stay in the niche markets of itch.io, peddling their £3 pdf games? Do they print out a few A5 stapled booklets and hope to get a few into distro with people like All Rolled Up? Do they make the leap of faith and go all in, like Rowan, Rook and Decard?

Which leads me to

3. In the same way that Print on Demand in the mid 00s changed the game for indie designers, by providing access to a different form factor of production at a much more accessible price point - and thus, for a time changing the look and feel of RPGs - AI art is going to do the same. It provides a different form factor in terms of art. Where the issue lies is that where POD took a chunk out of the lifestyles of traditional printers, AI art takes a swipe out of artists, who might well be people we know.

OR

We, as RPG consumers, will start to signal to the market that we are happy once again with basic art/layout games at lower price points, and we will see an art-lite revolution, leaving the big beautiful books to the professionals and letting the indie designers loose once again without the perceived constraints of $300 per page art requirements ... or the shackles of AI protest.
 

torus

Rune Priest
#12
Good post. Very much hope it's the second of those outcomes. And I'd really encourage rpg designers to learn more about layout and typography, because it's perfectly possible to make beautiful books with less art. Indeed many people currently make ugly books with lots of art, because they ignore the importance of whitespace, typface selection and the basic principles of typesetting.
 
Last edited:
#14
I'm all for conciseness - many games/scenarios are overwritten, in my opinion.

Going back to @Vodkashok's post, this particular hobbyist was very grateful for Arflow.ai's portrait generator (now "Legacy Image Studio") and its ability to tweak portraits for my for Other London books. For other art, I used photos from Pexels and Pixmania and played around with them in a graphics package.

The books would have been much duller without the art. (And while I am not sure they are beautiful, hopefully they're not too ugly.)
 

torus

Rune Priest
#19
I think Modiphius were the first to bring us waffle in the form of art, in their star trek books, which contain so much visual clutter and random graphic cruft that on some pages it's hard to work out what if anything is the actual content.
 
Top