Mod+ Ai generated content in RPGs

Best Selling RPGs - Available Now @ DriveThruRPG.com
I might give Night Cafe another try.

...

It feels good to me when I see pictures of women who look my age or older as adventurers.
Some efforts on Night Cafe:
fND21kPJ2tbJLTIvyovf--1--vheix.jpg

bFuv93u7ItTGP3hRVbH2--1--usntv.jpg

fTKgLTqxaPdCj5QfRvHA--4--9j7uf_2x.jpg

eUqBvcZFfBcgEXpbYeUh--1--o82a7.jpg

XIJG2SDPennMUNC4DENt--1--s7b0f.jpg

89TpIRCmftcjetrfUVCX--11--xx08l_2x.jpg

OaN21NHOttJlebpoF1vA--13--wvpsh.jpg

CmswwWisP551tmdG4z4Q--1--ipplv_2x.jpg

DkqJCV0YqPsB2M4gUEsm--2--siqnh_4x.jpg

KOloAhkQ7P270DWrycSD--2--1on0p.jpg

M2IHsufr7gOxcBT20JOd--1--5c195.jpg

M2IHsufr7gOxcBT20JOd--2--5dclo.jpg

wBgjqbOubcMifipN9QnG--4--9v43t.jpg

qxuv1oJXy6RNV2cQKAis--1--afvkb.jpg

sZTKlVqZFcqRmOpLIxlG--1--9lt7g_2x.jpg

YqpfQOUDKHlMqPyUnrfK--2--p9tr8_2x.jpg
 
I was able to get generate some old women on night Cafe. I suspect that certain presets tend to skew more toward young and beautiful than others. However, since I can only do a few a day, I am not able to verify it with a meaningful sample set.

A few days ago, I used the Fantasy preset and got young women with white hair.

Yesterday, I used the default setting and added the word Fantasy to the prompt. I got this set of warrior women:

ZEGUPWahV3qN7W2AN33w--3--ppnmv.jpgZEGUPWahV3qN7W2AN33w--2--exhlg.jpgZEGUPWahV3qN7W2AN33w--1--rc7lk.jpgZEGUPWahV3qN7W2AN33w--4--68y57.jpg


And for a very different vibe, I have been playing with the prompt "magical librarian". None of them look as magical as I wanted, but I think they turned out pretty good. If you don't look close at some of the hands

yjVF1Bh3y8Fhu4Genop2--4--fcnow.jpgyjVF1Bh3y8Fhu4Genop2--3--a7r9x.jpgyjVF1Bh3y8Fhu4Genop2--2--fdefk.jpgyjVF1Bh3y8Fhu4Genop2--1--yccyn.jpg
 
And for a very different vibe, I have been playing with the prompt "magical librarian". None of them look as magical as I wanted, but I think they turned out pretty good. If you don't look close at some of the hands
Arthritis and the occasional kitchen knife slip can screw up old hands pretty bad. The librarian hands look reasonable on a phone screen as long as I don't zoom in so far that they're the major focus.
 
Arthritis and the occasional kitchen knife slip can screw up old hands pretty bad. The librarian hands look reasonable on a phone screen as long as I don't zoom in so far that they're the major focus.
I am doing these on my phone so I have the same issue. On a phone screen, some mistakes are less noticeable.

I think that is both good and bad. I enjoy the pictures more since the flaws are less noticeable. On the other hand, I will probably post stuff online that has mistakes everyone but me will notice.

I have been doing some post apocalypse fighters and thinking maybe the strange hands would work. They are MUTANT Crall Classic characters, of course they have mutated hands.
 
Occasionally I see people online say that all AI art looks the same. Next time I read that, I may respond with my Appalachian possum shaman.
People who say that don't know how to write good prompts, or they're using the wrong software. I've seen a lot of really cool and interesting AI art. Awesome piece, by the way.

I think where AI art misses the mark is when you're looking for something pretty specific. AI art can look good, but it's hard to realize a complex description the way you want it to look.

Even though I'm totally fine with AI art, for the adventure I'm working on for publication, I'm sticking exclusively with human artists. I have some very specific styles and scenes in mind, and my many experiments with Midjourney have fallen very short of what I'd like to see.

Speaking of which, here's the placeholder cover image I generated for that adventure:

BKCC - cover.jpg

Not bad considering that I requested a combination of the styles of Frank Franzetta and Moebius! I mean, I can really see both influences.
 
People who say that don't know how to write good prompts, or they're using the wrong software. I've seen a lot of really cool and interesting AI art. Awesome piece, by the way.

I think where AI art misses the mark is when you're looking for something pretty specific. AI art can look good, but it's hard to realize a complex description the way you want it to look.

Even though I'm totally fine with AI art, for the adventure I'm working on for publication, I'm sticking exclusively with human artists. I have some very specific styles and scenes in mind, and my many experiments with Midjourney have fallen very short of what I'd like to see.

Speaking of which, here's the placeholder cover image I generated for that adventure:

View attachment 77538

Not bad considering that I requested a combination of the styles of Frank Franzetta and Moebius! I mean, I can really see both influences.
In most of my attempts, I am not looking for anything specific. So I have a decent chance of getting something I like.

But sometimes I get something nothing like the prompt. The styles seem to be especially bad for this. I recently tried the prompt "old punk lady" and the comic style. It gave me costumed superheroes that did not look remotely punk. Or old.

Umm...not all comics are costumed superheroes. Punk indy comics are a thing.
 
So for any of you who are struggling with the problems of preset styles, I noticed I get results closer to what I want by typing the style in the prompt and using the default style.

Example (with the starryai app):

Prompt: old punk lady.
Style selected: comic book
Result:

starryai_8wusb.png

Prompt: old punk lady. Comic book style.
Style selected: default setting

Result:


S


starryai_5e81n.png
 
So for any of you who are struggling with the problems of preset styles, I noticed I get results closer to what I want by typing the style in the prompt and using the default style.

Example (with the starryai app):

Prompt: old punk lady.
Style selected: comic book
Result:

View attachment 77554

Prompt: old punk lady. Comic book style.
Style selected: default setting

Result:


S


View attachment 77555
Google Bard response to:
Draw an old punk lady in comic book style
 

Attachments

  • Bard_Generated_Image (1).jpeg
    Bard_Generated_Image (1).jpeg
    588.4 KB · Views: 3
  • Bard_Generated_Image (2).jpeg
    Bard_Generated_Image (2).jpeg
    435.8 KB · Views: 3
Nice work. What art style generated this
 
Why does she have a moustache?

It's not a moustache, if you enlarge the pic it's clear that's her upper lip. But she has a weird mouth anyway and her hand is also weird. Well AI at it's best I guess.
 
The entire debate over generative AI is tiresome, mainly because the technology is here, will be around for the foreseeable future, and will only improve. So, almost any criticism like they can't draw hands is constantly negated by now. And that's a perfect example because Midjourney (for one) used to be terrible at hands but now rarely makes mistakes.

Whether it's "good enough" is a wholly subjective discussion. It's certainly good enough for my purpose, which is to do amateur design work. I also use it to make things like banner ads for proposed games on my Discord server. In this case, it doesn't have to pass professional scrutiny; it just has to look good to me and my players.

In the end, generative AI is like any other technology. You'll either find it helpful or you won't. If you don't, someone else will. And that's okay.
 

Attachments

  • Sands of Allegiance.jpg
    Sands of Allegiance.jpg
    48.7 KB · Views: 6
The entire debate over generative AI is tiresome, mainly because the technology is here, will be around for the foreseeable future, and will only improve. So, almost any criticism like they can't draw hands is constantly negated by now. And that's a perfect example because Midjourney (for one) used to be terrible at hands but now rarely makes mistakes.

Whether it's "good enough" is a wholly subjective discussion. It's certainly good enough for my purpose, which is to do amateur design work. I also use it to make things like banner ads for proposed games on my Discord server. In this case, it doesn't have to pass professional scrutiny; it just has to look good to me and my players.

In the end, generative AI is like any other technology. You'll either find it helpful or you won't. If you don't, someone else will. And that's okay.

Your right, the technology is here and will not go away. There's also nothing wrong with the way you are using it. If the subscription cost for something like Midjourney wasn't as high as it is, I would probably have fooled around with it myself.

But as this thread has already covered, there's the whole issue about copyright/ownership of AI generated stuff. As well as the issue about how the companies behind these programs, generate their databases to use for the AI.
Of course this is not relevant to the way you use it.
 
It's adjusted AI - like they start with a generated image and then go in and tweak it.
Wow... I wonder how much adjusting it took? Either way it looks fantastic. And to think that it can be produced so quickly too.

I think I'll be out of a day job soon, lol
 
The entire debate over generative AI is tiresome, mainly because the technology is here, will be around for the foreseeable future, and will only improve. So, almost any criticism like they can't draw hands is constantly negated by now. And that's a perfect example because Midjourney (for one) used to be terrible at hands but now rarely makes mistakes.

Whether it's "good enough" is a wholly subjective discussion. It's certainly good enough for my purpose, which is to do amateur design work. I also use it to make things like banner ads for proposed games on my Discord server. In this case, it doesn't have to pass professional scrutiny; it just has to look good to me and my players.

In the end, generative AI is like any other technology. You'll either find it helpful or you won't. If you don't, someone else will. And that's okay.
I think the genie is out of the bottle and it won't ever go away... The tech is improving exponentially as well.

What I hope will happen is that hand drawing (even digitally) is still valued as a craft and that AI is seen as the cheap and easy way route to go. Basically like having a big mac as opposed to one that's been made by a chef.

Being a bit of a lefty and a cynic I doubt many of the bigger companies will really care and the majority will eventually go AI (when the hubbub dies down of course). They tend to be more concerned with turnaround speeds and costs. :sad:

I suppose I better get good at key wording. lol
 
Generative AI is going to be a substantial part of all our work lives, I think. I just hope that it gets sensibly regulated - you should be required to license your training sets, for example, and it should be completely illegal to use any kind of machine learning model to make decisions about, e.g. loan approvals and the like. We shall see.

More on topic. I think it's clear that companies will start allowing AI, apart from anything else, it's going to become part of the normal workflow of most digital artists. Maybe not the prompted image generation side of thing so much, but getting AI to fill in the background when you move something or want to expand the canvas and so on are just going to become completely normal and the line between "generative art" and "using the AI to clean up a bit" is going to be completely blurred.
 
Banner: The best cosmic horror & Cthulhu Mythos @ DriveThruRPG.com
Back
Top