A.I. Imaging

Best Selling RPGs - Available Now @ DriveThruRPG.com
Status
Not open for further replies.
I'm pretty open to it... I like the fact that it can give creators a voice if you will for their own creations that they would not have been able to have before.
I would certainly see it as an extra tool in the 'tool box' so to speak.
The hysteria around it has indie game producers saying that you're a bad person to use it in your games and they'll be vocal about it.

I don't think it's a bad thing any more than any automation has been a bad thing but I find this claim of harm to be .... homeopathic violence. That's the phrase.
 
Not sure I want to make it a habit for anything we release, but I wanted to get a shot of Nazi's invading Santa's Workshop and that's a tall order in the time period that I had for our release schedule (about a week). My brother put a pair of images together for me using Dali, and we disclose how they were generated in the credits section. The adventure is PWYW/free, and it's not something I plan to do for any larger projects.

After seeing what they were able to do I started playing around in Night Café for my home game. All of these have been for personal use only.

1670773314125.png

1670773387777.png

1670773422897.png
 
I'm equally excited about chatgtp. I keep reading how high school and humanities teachers are saying it's the end of writing as a filter. Not all agree it's a bad thing. To me I see it a bit like the cheap calculator. Prior to that being invented there were a whole lot of people excluded from dealing with mathematics because either they had issues with math or slow with it. After they were allowed into things like engineering and expanded the pool of people able to solve problems. How many more RPGs are we going to get now that cover a niche due to chatGTP and midjourney? I suspect quite a few.
 
I'm equally excited about chatgtp. I keep reading how high school and humanities teachers are saying it's the end of writing as a filter. Not all agree it's a bad thing. To me I see it a bit like the cheap calculator. Prior to that being invented there were a whole lot of people excluded from dealing with mathematics because either they had issues with math or slow with it. After they were allowed into things like engineering and expanded the pool of people able to solve problems. How many more RPGs are we going to get now that cover a niche due to chatGTP and midjourney? I suspect quite a few.
How do you see chatGTP contributing to RPGs?
 
How do you see chatGTP contributing to RPGs?
I did another thread here recently where I asked it to come up with a background for a sci Fi universe. It's not fantastic but as text to start with and work around it's not horrible.
 
Interesting development. chaosium says no. Despite most AI art being perfect for it. (The hands! The hands!)

interesting wording too. Not even artists are permitted.

(are they going to ban the “heal/retouch” function?)


 
Interesting development. chaosium says no. Despite most AI art being perfect for it. (The hands! The hands!)

interesting wording too. Not even artists are permitted.

Ooh, a rebellion is brewing then. How many will follow suit? Will it achieve anything?

The history of Luddism isn't exactly replete with success, but who knows, maybe things will turn out differently now that as a culture we're effectively building soft social credit systems using internet media.

Cancelled for using textile machinery!

ucPFbp6.png


Still, I wonder how exactly they're supposed to police this. Do they insist that all artists video-record the drawing process? What if they used AI to create an image they then hand-copy from somewhere out of frame? :tongue:

MVDUr5I.gif
 
Last edited:
Interesting development. chaosium says no. Despite most AI art being perfect for it. (The hands! The hands!)

interesting wording too. Not even artists are permitted.

(are they going to ban the “heal/retouch” function?)



This reminds me of the time I think I mentioned earlier in this thread, where a guy wanted me to create a logo for them, but wanted me to draw it from scratch and not use premade icons as a base template to work from. There's a line between paying me to produce something and telling how I'm allowed to produce it, and this crosses it. Giving me the specs is obviously a requirement to achieve the desired outcome, but telling me what tools I'm allowed to use isn't giving me specs, it's telling me how to do my job.
 
I'm going to release my first product next week that includes A.I. generated art. DriveThru's latest policy is that you have to designate your product as containing A.I. art during upload and then another disclaimer somewhere else... I think in the product's description. Apparently, customers demanded it.

Anyway, Kort'thalis Publishing will be proudly presenting Encounter Critical III. I'm excited to see what you guys think of the text and art! Personally, I think a good mix of organic and machine art will be the way to go for the next few years. Keeps things economical and visually stimulating.

VS
 
I'm going to release my first product next week that includes A.I. generated art. DriveThru's latest policy is that you have to designate your product as containing A.I. art during upload and then another disclaimer somewhere else... I think in the product's description. Apparently, customers demanded it.

Anyway, Kort'thalis Publishing will be proudly presenting Encounter Critical III. I'm excited to see what you guys think of the text and art! Personally, I think a good mix of organic and machine art will be the way to go for the next few years. Keeps things economical and visually stimulating.

VS
What's the penalty for failing to designate correctly?
 
I am a computer person by trade who was an artist by training (a long time ago, though I doodle from time to time). I just see it as another tool, to be used or not as needed or desired. I am indeed amazed at how well it works and how I imagine it will get even better over time. I don't care how the art was made, I just care that it fits (or sets) the tone of the product and adds to the experience of reading or playing it.
 
And how will they be able to known? Does AI art leave some sort of imprint in the back end they can look for? Or will they just be counting fingers on any hands in the art?

This is just silly.
AI Art without human clean up has problem with context. The model is aware of patterns but not why they exist, so you will have things like mismatched ears, hair not the same on side as it is other, stuff that looks human made like art on a piece of jewelry and so on. These are not gross errors but they are easily spotted if you know what the look for. And incidentally are the kind of things a human artist would be criticized for not getting right.

You can see what I mean here

As well as explain how to do it.
 
There is probably nothing that humans can do that AI will not be able to do better eventually. It's what we decide to do at that point that really matters. In my own former industry (music), this started in the 80's and no one did anything about it because there's nothing to be done. Virtually no one cares how music is made or if it was done with skill or creativity. Music is a dead art form already as far as popular content goes, and has been for some time. Visual art will go the same way, as will all human endeavours. Chess is another prime example. Why sculpt when you can 3D print, etc.
 
Part of me really wants to kick against AI art because of the way it has come about. Part of me acknowledges the coming singularity and thinks there might be bigger hills to die on. Also I've never given a feck about art in RPGs. Peaked with 1st ed AD&D, if you ask me.

Some artists are leaning into it as a new medium, however, and these will most likely be the artistic overlords of the coming new era. Check out Sentient Muppets and Bland Space on Instagram if you want to see some real-freaky-cool AI art.
 
I've found AI art decent at background/landscapes, but not so good when you want a specific figure. Combining stock art with AI backgrounds makes for some nice combinations, the best of both worlds as it were.

For example, some art by Jeremy Hart I combined with an AI background (modern city, giant spider webs, overcast, comic style).

zzEVY1e.jpg



I've yet to use anything but commissioned and stock art in my published work, but at some point some AI will probably get used. I do think artists must be compensated if their art is used to train the AI. Any corporation will happily automate any job (not upper management of course) to save money and while they have come for the artists right now, most everyone is a target.
 
I've used AI art for 'trim' and filler. But commissioned art and stock images for the actual important stuff.

The whole argument is BS though - even with the "omg it was trained on my art" claim. The "harm" is in their heads even if they had it trained on their art.

In a decade it will be an essential tool for an artist. No-one expects them to design their own gradients. They use heal and retouch with impunity. It will just be a labour saving device.

THAT SAID...

It should be easier for artists to train their own models. I think it should be retasked as a tool for artists.
 
AI Art without human clean up has problem with context. The model is aware of patterns but not why they exist, so you will have things like mismatched ears, hair not the same on side as it is other, stuff that looks human made like art on a piece of jewelry and so on. These are not gross errors but they are easily spotted if you know what the look for. And incidentally are the kind of things a human artist would be criticized for not getting right.

You can see what I mean here

As well as explain how to do it.

Yeah, I assume that a human artist will have to retouch the image--maybe even rework it substantially depending on the specs for what they want. But once a human has gone over it and cleaned up any weird AI smudges, how will they know? And what does it matter at the end of the day, if the illustration depicts what it's supposed to depict?

I do get the concern of using a human artist's art to train an AI, since it's ultimately using the artist's own art to effectively screw them over to some extend. But on the other hand, do I owe an artist royalties if I use images of their art as reference and use their style as inspiration while practicing my own art skills? IDK.
 
I suspect that DTRPG just put the disclaimer in to avoid getting dragged into the drama around AI art.
 
There is probably nothing that humans can do that AI will not be able to do better eventually. It's what we decide to do at that point that really matters. In my own former industry (music), this started in the 80's and no one did anything about it because there's nothing to be done. Virtually no one cares how music is made or if it was done with skill or creativity. Music is a dead art form already as far as popular content goes, and has been for some time. Visual art will go the same way, as will all human endeavours. Chess is another prime example. Why sculpt when you can 3D print, etc.
The better part I don't necessarily agree. Good enough I agree. Given they can do it faster, cheaper, on demand and "good enough" means a ton for people. It's such a drop in barrier to entry. I can now do art. Orders of magnitude better than I actually physically can. Good enough for me. That didn't exist last year.
 
I've been playing about with AI Art since this thread started. Creating the art I want, as it turns out, is an art all in itself. Somehow the first image is always the closest to what I want, but never quite gets there. Then you spend hours down the rabbit hole trying to tweak it. The AI seems to excel at doing dystopian cityscapes, but most of the stuff I've been trying to squeeze out of it, while decent, isn't what I wanted or just 'off'.

If your game is on the bizarre end of the spectrum I feel AI art might be more useful. I can tell the AI to create an old brickwork cellar with faces, long shadows, and twisted swirly walls, but it doesn't seem to have a clue when I want wine barrels along the walls to make it look like a wine cellar. Ask for eyes and you get them everywhere but in the person's head or they appear as just bright points in random places.

I haven't really formed an opinion yet if it's good or bad for artists, it's certainly eating up a lot of my spare time. Recently we got a new IT system at work. It's supposed to have all the bells and whistles and be more efficient, but in reality, the bells and whistles are only nice to have if you are the one person that uses them, they get in everybody else's way. It's a bigger database and consumes more data that still has to be entered by actual people. Now we need more people. So, like the AI, I think there is a long way to go before people start losing jobs en masse and folks will have moved on before that ever happens.
 
I do get the concern of using a human artist's art to train an AI, since it's ultimately using the artist's own art to effectively screw them over to some extend. But on the other hand, do I owe an artist royalties if I use images of their art as reference and use their style as inspiration while practicing my own art skills? IDK.
So morally human art students shouldn’t study contemporary art to avoid using another artist’s art to effectively screw them over by copying their technique and themes?

Legally it OK for this, human art students, to happen because the artist only has the legal right to control the copying of a specific image. Also some countries recognize a moral right for the artist work to remain as is in the form the artist intended. But that doesn’t prevent the copying of themes and techniques.

Just trying to see where the line being drawn here. As the AI models do not store the actual images themselves.
 
I can see several threads in the current debate about this. There are probably more, but this is how I can see it affecting me.

One is corporate use of copyrighted material illegally hosted online to train their AIs. A case could be made that there should be a legal framework to prevent this, it would be easy to restrict AIs to only archives of legally copyright-free or specifically permitted material. On the other hand, the internet is the internet and trying to control what people do with your stuff once it's out there is a fool's errand.

Two is the impact on artists' livelihood. This is sad, but inevitable. Any effort spent to try to stem the tide would be better spent in retraining or specialising in particular niches which will still be viable once mass-produced AI art becomes truly viable. The Arts & Crafts Movement did nothing to reverse the Industrial Revolution, but it did create a whole high-value (expensive) alternative market to mass production. Only a tiny minority of artists would be able to benefit from this, and only a small proportion of clients would be able to afford them (certainly not RPG publishers). Remember that crap they told us in school how robotocisation of factories would put factory workers out of their jobs, but they would open up other industries to replace them? Yeah, that never happens or, if it does, it's a downgrading because humans are pushed into less skilled lower-income work.

Three is the democratisation of art. Like mass-production, AI art (once it works) will put bespoke art in the reach of people who would never be able to afford individual commissions or original art. RPGs are limited to the bottom tier of artists in terms of price (not ability!), and most could never afford the type of high-end full-colour work that AIs will (probably) soon be capable of. This is not a case of artists losing clients, this is publishers using AI instead of public domain or licensed clip art or no art at all. Full-colour covers are a major expense that take a long time to recover and are just unviable for small, cheap, or esoteric books. I know that my entire publishing budget goes on artists, I'm doing the writing, editing, and layout for free. I'm just offsetting the cost of my hobby, really.

Four is the raising of expectations. Soon, cheap and cheerful will no longer be acceptable even for indy RPGs with sales in the low hundreds. Just like games like D&D 3E pushed other professional RPG companies to go full-colour even though their sales were just a fraction of WotC's. It will increase the tendency towards homogenisation, because necessity is often the mother of innovative design.

The bottom line, though, is that it doesn't matter how any of us feel about it, because it's far too late to stop it. Humans as a society haven't been in control of progress for a long, long time, all we can do is try to stay on the train and not get run over.
 
Last edited:
I've been playing about with AI Art since this thread started. Creating the art I want, as it turns out, is an art all in itself. Somehow the first image is always the closest to what I want, but never quite gets there. Then you spend hours down the rabbit hole trying to tweak it. The AI seems to excel at doing dystopian cityscapes, but most of the stuff I've been trying to squeeze out of it, while decent, isn't what I wanted or just 'off'.

If your game is on the bizarre end of the spectrum I feel AI art might be more useful. I can tell the AI to create an old brickwork cellar with faces, long shadows, and twisted swirly walls, but it doesn't seem to have a clue when I want wine barrels along the walls to make it look like a wine cellar. Ask for eyes and you get them everywhere but in the person's head or they appear as just bright points in random places.
That was my experience when I played around heavily with midjourney a few months ago. Totally rocked city stuff, but something as "simple" as a shark or airplane or even a superhero, was out of the question. I believe it has to do with symmetry, or lack there of. I posted examples in other threads, but here is a decent one of a pirate outpost done as a watercolor:

Picaroon_Jack_watercolor_painting_caribbean_shanty_town_outpost_d694e1ff-df8f-4cd4-82c3-913e75...png

The fact that it's a shanty town made of old sails made it even easier to avoid the symmetrical issues.
 
I'm sure DTRPG will be happy to work with publishers to get them to designate correctly. It's not like they want to lose their business.
 
Yeah, I assume that a human artist will have to retouch the image--maybe even rework it substantially depending on the specs for what they want. But once a human has gone over it and cleaned up any weird AI smudges, how will they know? And what does it matter at the end of the day, if the illustration depicts what it's supposed to depict?

I do get the concern of using a human artist's art to train an AI, since it's ultimately using the artist's own art to effectively screw them over to some extend. But on the other hand, do I owe an artist royalties if I use images of their art as reference and use their style as inspiration while practicing my own art skills? IDK.

Isn't that what a novice human artist does? Find one or several artists and copy their style until they master it to some degree and broaden their aesthetic horizons until they create their own individual style, standing on the shoulders of giants.
 
Isn't that what a novice human artist does? Find one or several artists and copy their style until they master it to some degree and broaden their aesthetic horizons until they create their own individual style, standing on the shoulders of giants.
Well, yes, but these artists ask permission and pay copyright dues.

No. Wait. They don’t.

It’s exactly what they do and what they have been doing for centuries. But it sucks when a machine does it quicker.

meanwhile … farm workers, supermarket workers, print workers and looking on and thinking “ya”
 
...

If your game is on the bizarre end of the spectrum I feel AI art might be more useful. I can tell the AI to create an old brickwork cellar with faces, long shadows, and twisted swirly walls, but it doesn't seem to have a clue when I want wine barrels along the walls to make it look like a wine cellar. Ask for eyes and you get them everywhere but in the person's head or they appear as just bright points in random places.
...
I've had the same experience myself, mostly. However, I just managed to get these in two passes. I don't know if you really wanted these, but in case:
0LumI62jl8wUuKK2awTc--1--5m6t0.jpg
0LumI62jl8wUuKK2awTc--2--qjj83.jpg
0LumI62jl8wUuKK2awTc--3--9jgf0.jpg
0LumI62jl8wUuKK2awTc--4--bqknf.jpg
I used "an old brickwork wine cellar with long shadows and twisted swirly walls, with wine barrels along the walls" in Stable Diffusion v1.5 with Prompt Weight 90%, which gave me four views of brick halls with curved ceilings and no barrels at all. Then I fed the most "wine-cellary" one back in as a seed image, with the only prompt "a row of wine barrels along the walls", and it gave me the above images. I was shocked, as I rarely get such obedient results.
 
With this many people playing with it, finding failings and reporting them back either in email or just by what they click on or not I would expect a lot of rapid improvement from the various systems.
 
The AI seems to excel at doing dystopian cityscapes, but most of the stuff I've been trying to squeeze out of it, while decent, isn't what I wanted or just 'off'.
I've been finding them OKish at single "thing" images & prompts where there's probably a decent body of existing work, sometimes with a combo but sometimes not. And yeah, there's typically issues with symmetry.

Caveat: I'm not trying too hard, just 20-30 minutes of alternating prompts & styles on free online ones. Alas, I don't get much actual computer time outside of work to do anything but some text & light spreadsheet stuff.

Like any combo of Elvis and Godzilla basically didn't work. You got one or the other. Not even for prompts about Elvis riding to top of Godzilla. But after a few tries I could get 50/50 decent zombie pirates or zombie rockers in Metallica style.
 
What's the penalty for failing to designate correctly?

The full guidelines are described here, but in short:

1. If you use AI art in your product, then you have to check the "AI-Generated" search filter under "Format", and:

2. If you sell AI-generated stock art, you also have to include a disclaimer in the product description.

The guidelines state "Titles that do not comply are subject to removal from the marketplace. Repeat offenders may have their publishing permissions revoked."

However, I know someone is going through the products updating the DTRPG filters, so realistically I think they'd just update your product for you, unless it's clear that you're intentionally "forgetting" (or worse, removing the filter after they've set it).
 
Status
Not open for further replies.
Banner: The best cosmic horror & Cthulhu Mythos @ DriveThruRPG.com
Back
Top