Viridian's AI Experiments

Artificial Intelligence is here! Really! Anything created with AI assistance, including stories, should be posted here.
User avatar
MadMax359
Posts: 4214
Joined: Sun Apr 19, 2009 5:07 am
Location: north carolina

Re: Viridian's AI Experiments

Postby MadMax359 » Sun Dec 11, 2022 8:40 pm

it was worth asking... and i think Quicky in the mine worked great--- in the end, the supporting details are secondary to the damsel in distress--- but i know you have standards!
The strong do what they want, the weak do what they must

Viridian
Posts: 1587
Joined: Wed Apr 15, 2009 10:03 am

Re: Viridian's AI Experiments

Postby Viridian » Mon Dec 12, 2022 6:32 am

Acidtester wrote:I’m not a huge fan of AI “art”, but some of these come out okay.
I have a question tho, can AI not figure out how to make a scared face? All of these girls look bored, like drowning to death in quicksand is on their to-do list, but it ain’t their whole day.

Part of the cause is that my reference images - my "studies" - use neutral model expressions. Consequently, the AI adapts those expressions, which for the purpose of what I was experimenting for, was sufficient. It's possible to input preferred tags, such as "scared" or "concerned", and the AI draws upon its database of artistic expressions which match them. Often they don't really match what we expect, and part of it is because attach a lot of subjectivity to the, well, subject matter. A "scared" face might not match as well as a "shocked" or "surprised" or "disgusted" look in the same context. You can go even further to specify exact objective expressions, such as a slight grin, or squinted eyes, or blushing cheeks. The challenge in generating good "art" is that you're not using the pen. You're understanding what to plug into the machine to get the desired result.

Plus, having excellent reference images crucial. The closer the image is to the desired output, the better and more consistent the result. That means not having to change colours as much (such as changing clothing colours or big hair colour changes, like red to blonde), the right facial expressions, the right quicksand texture, etc. What I'm doing now is saving a lot of my "failed" experiments because they contain a lot of useful references, such as clothing, texture and angles, to adapt into other pieces.

As an aside, I tried using your Ruby Roundhouse artwork to see what the machine could spit out. Might not be your thing, but here's what I got.
ruby1.png

ruby2.png

ruby3.png

ruby4.png


Believe me, it took dozens of attempts too. I can't recreate the exact costume without being uber-specific on what she is wearing, and the more you throw in, the more it confuses the AI because it means you have to be even more specific, and the AI will still be liberal in interpreting the possibilities, leading to more branches. I was never going to attempt to include the pack strap because it will never get it exactly in the same place and alignment, whereas as the artist you just paint the thing on. The gloves were interesting too. I started with "black gloves", and a bunch of images threw up bulky military combat gloves. "Black leather" worked better, but then I remembered that Ruby has FINGERLESS gloves, so that got put in too to generate these. That's really the battle (or the "skill") with AI "art".

MadMax359 wrote:it was worth asking... and i think Quicky in the mine worked great--- in the end, the supporting details are secondary to the damsel in distress--- but i know you have standards!

Yeah, well, what it did generate was pretty decent. The art style, alone, is very nice. It's that it took about 50 attempts and a lot of fine tuning because the AI doesn't recognise what the scene is. For us, visually, we know it's a character strung up by a pipe with her thighs sinking into quicksand with her skirt slipping, her blouse ripped open, and her breasts popping out of her bra. But convey that to a machine and they like "So, like, her arms are stuck in mud in the ceiling and she has three legs, right?"

It also didn't help that my artwork wasn't tailored to get a good read for the AI. It was too dark, with too many details that we can piece together, but it's all noise for the AI. It was almost impossible to get something resembling mud or quicksand on the floor because, frankly, it ISN'T mud/quicksand. Your brain is telling you that it is because of context. But it isn't, and the AI struggles to piece that together clearly. Tag it enough and you can nudge it close enough in the right direction to get SOMETHING that resembles it. Ultimately, what I generated was a "weak" version - that is, a high "strength" generation allows the AI to recreate the concept more freely, while a "weak" generation stays more close to the original. While this is good for keeping the construction of the piece, it can fail artistically because it's not what you want from the AI.
You do not have the required permissions to view the files attached to this post.
Viridian @ deviantART: http://viridianqs.deviantart.com/

Viridian
Posts: 1587
Joined: Wed Apr 15, 2009 10:03 am

Re: Viridian's AI Experiments

Postby Viridian » Mon Dec 12, 2022 12:33 pm

So here's something different. It's easy to spam out decent-looking concept art, but that's a cheap way to create art. While creating quicksand scenes is incredibly challenging, generating other stuff is not. The final frame took a whole day to figure out. The AI simply cannot produce it on its own. I fed it dozens of my samples and I couldn't get exactly the right pose and texture. I might have been able to brute force it if I refined my parameters more, and in hindsight I could've done this more easily know that I know what the settings do. I ended up drawing the base image myself - at least, a shitty 2-layer version of it so that the AI can pick up what the heck I was trying to show. Ran that one through a dozen iterations to get the final picture.

But to show what the AI can do, I've attached the "Fuck you, just gimme quicksand" sketch of the final frame. Warning, it's not pretty: https://sta.sh/0cu861je5bx
You do not have the required permissions to view the files attached to this post.
Viridian @ deviantART: http://viridianqs.deviantart.com/

User avatar
MadMax359
Posts: 4214
Joined: Sun Apr 19, 2009 5:07 am
Location: north carolina

Re: Viridian's AI Experiments

Postby MadMax359 » Mon Dec 12, 2022 2:15 pm

Business Class takes on a whole new meaning--- she has class and means business! :twisted:
The strong do what they want, the weak do what they must

Viridian
Posts: 1587
Joined: Wed Apr 15, 2009 10:03 am

Re: Viridian's AI Experiments

Postby Viridian » Mon Dec 12, 2022 4:30 pm

MadMax359 wrote:it was worth asking... and i think Quicky in the mine worked great--- in the end, the supporting details are secondary to the damsel in distress--- but i know you have standards!

Actually, I take that back. My work on the AI today has brought up the "right" way of making adjustments. There's an inbuilt image editor. It's a a very bad one with literally only one tool (the pen), but often that's all you need to paint over something to make the AI recognise what it's supposed to be. I've been able to add clothing, remove clothing, add holes, tie ropes, and most importantly, adjust the depth and texture of the quicksand using the equivalent of a Sharpie and letting the AI figure out that my squiggly lines are mud ripples.

And it works. I'm extremely pleased that Mine v2 captures the spirit of my drawing. The removal of the background makes it better, IMO. The addition of the darkness, plus the dripping water, really makes it feel that Sanders has been abandoned and is being lowered into the quicksand.

I also re-did my avatar.
You do not have the required permissions to view the files attached to this post.
Viridian @ deviantART: http://viridianqs.deviantart.com/

duuudeization
Posts: 644
Joined: Mon Aug 06, 2012 6:43 pm
Location: Florida

Re: Viridian's AI Experiments

Postby duuudeization » Mon Dec 12, 2022 5:39 pm

Kinda liking those last 3 entries! :D
Its all about peace and love and good happiness stuff

User avatar
sixgunzloaded
Posts: 918
Joined: Tue May 05, 2015 8:16 pm

Re: Viridian's AI Experiments

Postby sixgunzloaded » Mon Dec 12, 2022 7:30 pm

As someone who is into CGI in general, I gotta say, some of this stuff is really impressive. I know it's not really thought of as true 'art' in the creative sense, but the fact that a machine can interpret dialog and source images into new images is still pretty amazing. And I'd guess that there's still at least some creativity in how you input the parameters.
One could look at it as cheapening the art but, on the other hand, it does allow those who have minimal or zero talent to produce something that might be pleasing to look at. I just say that as someone who picked up Poser because I couldn't draw for squat, and I love producing stuff.
On the other other hand, it's also a little disturbing that Skynet now knows we're into quicksand. :P :lol:
How long did Tarzan watch before deciding to save Jill..?

User avatar
SinkerCutie
Posts: 943
Joined: Wed Jul 15, 2009 4:50 am

Re: Viridian's AI Experiments

Postby SinkerCutie » Mon Dec 12, 2022 11:05 pm

Viridian wrote:A few more that turned out really, really well.

Would anyone like to run requests?

How about those babysitter pics you did?

Viridian
Posts: 1587
Joined: Wed Apr 15, 2009 10:03 am

Re: Viridian's AI Experiments

Postby Viridian » Mon Dec 12, 2022 11:43 pm

MadMax359 wrote:Business Class takes on a whole new meaning--- she has class and means business! :twisted:

Personally, this is exactly the sort of story I love most, and one that hardly gets ticked exactly the way I want it. There's no official narrative - the magic of it is that you can piece together the details. But I wanted to show progression. In my mind, she is on a business trip and the plane crashes. She survives and has to find her way through the hostile jungle. Her business suit is poorly suited for the harsh environment. The humid jungle makes her ditch her jacket. A rain storm drenches her. The dense vegetation tears her clothing (and for some reason or another, she's had to remove her bra...). Finally, battered and torn, she runs into inescapable quicksand.

sixgunzloaded wrote:One could look at it as cheapening the art but, on the other hand, it does allow those who have minimal or zero talent to produce something that might be pleasing to look at. I just say that as someone who picked up Poser because I couldn't draw for squat, and I love producing stuff.

That's pretty much how I got into doing visual quicksand material. I was always a writer first, but writing presents some limitations in telling stories for what is pretty much a one-shot genre (we want QS scenes, but it's hard to put together one really good quicksand scene or squeeze in multiple, and there's a mass consumption level). I wanted to see my ideas visualised, but no one was producing the exact content I wanted, so I started drawing them myself. I had no clue how to use Poser, so I went the digital art route.

I find I'm using AI the way others use Poser. My initial thoughts were tainted by seeing bad AI gens, similar to bad photomanips that just take a hot model, crop her legs and paste her onto a Studio588 set, or bad Posers where there is no interaction between the model and the supposedly deep quicksand. But just as there are good manips and 3D sequences (such as yours), after seeing someone else generate some fairly authentic sequences (albeit with further touch-ups), I really had to see if this was a medium that I could work with. Over the past few days, I'm growing a closer attachment to it. It fits exactly into how I saw myself: as a storyteller with a way to conceptualise my narrative.

There's certainly a high level of patience involved in getting the desired result. Given how easy it is to replicate good results, I feel that spamming out gens is exactly why people hate AI work - it's generic, uncanny and low effort. I think where I'm heading is to put together things that are more unique to what I can do rather than spit out anything that looks like quicksand. You're in it for the story; the visuals are a bonus.

SinkerCutie wrote:How about those babysitter pics you did?

Probably not going to happen. As I said above, these pictures are too complicated to generate with AI. It struggles to handle differences with two or more characters. I've been able to generate fairly generic scenes with two or even three. It may be doing to generate a scene where there are two generic similar characters where you don't care about what they look like, but there's too much human creativity to push out the legitimate art pieces we put together.

It's why I haven't jumped at the chance to generate Quicky Sanders material. Most of the scenes involve multiple character interactions with visually different appearances. There's no way for me to reliably generate a scene with one character with long black hair, glasses and red blouse up to her chest while a long blonde haired woman with a light blue tank top is being held at gunpoint by a masked man. That sort of thing. That's the sort of thing Poser can do, but while the AI can generate nice visuals, it doesn't generate good interpretations of ideas.
Viridian @ deviantART: http://viridianqs.deviantart.com/

User avatar
MadMax359
Posts: 4214
Joined: Sun Apr 19, 2009 5:07 am
Location: north carolina

Re: Viridian's AI Experiments

Postby MadMax359 » Tue Dec 13, 2022 6:30 am

i'm SO glad you kept on working at this... there's the Quicky we all love, and there are so many adventures she's had... :twisted:
The strong do what they want, the weak do what they must


Return to “AI Creations”

Who is online

Users browsing this forum: No registered users and 0 guests