• This is a political forum that is non-biased/non-partisan and treats every person's position on topics equally. This debate forum is not aligned to any political party. In today's politics, many ideas are split between and even within all the political parties. Often we find ourselves agreeing on one platform but some topics break our mold. We are here to discuss them in a civil political debate. If this is your first visit to our political forums, be sure to check out the RULES. Registering for debate politics is necessary before posting. Register today to participate - it's free!

AI generated images

ASHES

An Uncertain Person
Dungeon Master
DP Veteran
Joined
Sep 10, 2020
Messages
28,497
Reaction score
23,043
Location
Canada
Gender
Undisclosed
Political Leaning
Liberal
I won't say AI art in the title, that is an ongoing debate. I think it is and isn't, like any medium. It sure takes a lot more time and effort to get a great image, especially if you're going for something specific, than I thought it would. Seems easy enough, you just type in what you want to see, but then next thing you know you're learning about LoRAs, Weights, Checkpoints, Styles, Artists, Models, ControlNet, img2img, inpainting, the list goes on and on and on. I'm using Stable Diffusion 1.5 with Automatic1111 Web-gui. It runs locally on my PC and I use it from my phone on a web browser.

Here's a simple free online no BS signup/tokens one you can play with (potential for NSFW):

What can you make? (Let's keep it SFW/PG-13)


IMG_4693.jpeg

IMG_4691.jpeg

IMG_4695.jpeg
 
I've used AI images for work and played with it for fun.

"Ducks on the beach playing volleyball in the style of Monet" and goofy stuff like that.
 
I've used AI images for work and played with it for fun.

"Ducks on the beach playing volleyball in the style of Monet" and goofy stuff like that.
It can be used for that. It has great entertainment value. But I think it's potential is also overlooked by many.

BOOOO! You give us this knowledge and then punish us so?
With great power knowledge comes great responsibility.

While I am generally against AI "art" using it for silly things you wouldn't have paid anyone for anyway it fine IMO.
If the work from a talented AI artist user does the job, why wouldn't it be paid for?
 
If the work from a talented AI artist user does the job, why wouldn't it be paid for?
It doesn’t matter. AI art only works at all because it scrapes up real art without compensation. There is a reason that the courts decided stuff made with AI can’t be copyrighted. It’s just plagiarism with extra steps.
 
It doesn’t matter. AI art only works at all because it scrapes up real art without compensation.
I think that's a gross mischaracterization of what it is doing. You can plagiarize using AI, but everything created by AI isn't plagiarism.

There is a reason that the courts decided stuff made with AI can’t be copyrighted. It’s just plagiarism with extra steps.
Because a human didn't create it. Like why photographs taken by animals can't be copyrighted.
 
Cool thread idea. Here's Bugs Bunny climbing Mount Everest :

LW1 - 1 (36).jpeg
 
I think that's a gross mischaracterization of what it is doing. You can plagiarize using AI, but everything created by AI isn't plagiarism.
It literally is. Why does AI art often produce signatures of artists and water marks?

It would be completely impossible for it to produce any image without the art data so it knows what art looks like. They used massive amounts of art without compensating the artists at all.

It NEEDS human art because all it can do is copy. That’s it.

There is actually a problem where newer models are getting WORSE because they are starting to take in AI art in their data scraping. Using AI art to train AI makes it devolve and get worse like inbreeding.

So it can’t function without the artists it’s replacing. But the artists who allowed it to exist and completely copy their work to begin with aren’t compensated at all.
 
Here's Mona Lisa skydiving.

LW1 - 1 (38).jpeg
 
It literally is. Why does AI art often produce signatures of artists and water marks?
It literally isn't. It does that because it associates those things with certain styles and image patterns because the data it was trained on had them. What you won't find is an actual signature from an actual artist. It isn't copy/pasting bits of images, it's producing random noise it thinks resembles what you're asking for, and if it thinks that thing is associated with a signature or watermark, it will create that too, even though it's nonsense.

It would be completely impossible for it to produce any image without the art data so it knows what art looks like. They used massive amounts of art without compensating the artists at all.
It uses art data (Looking at pictures and associating them with keywords), just like a human would, and it isn't plagiarism for a human to look at a bunch of paintings by an artist, and then reproduce that style with their own creative input.

It NEEDS human art because all it can do is copy. That’s it.
It needs data sure, but it isn't limited to art and artists. That is a very small and well defined subset of training.

There is actually a problem where newer models are getting WORSE because they are starting to take in AI art in their data scraping. Using AI art to train AI makes it devolve and get worse like inbreeding.
I don't think that's much of a concern. That's like saying it's going to get worse because it looked at paintings by Jackson Pollock.

So it can’t function without the artists it’s replacing. But the artists who allowed it to exist and completely copy their work to begin with aren’t compensated at all.
Compensated for what? And how do you compensate Vincent Van Gogh, for example?
 
Shit. It's Batman. How random is that?

This is my first experience with AI stuff like this. One of the Mona Lisas skydiving had an extra arm.

View attachment 67497584
I find that fascinating.
They often have extra arms and legs. Hands are hard, and commonly have 6 fingers or more. Negative prompts help some. You can get perfect hands and poses using controlnet. That lets you extrct poses and expressions from pictures, or make your own.

PS: I added Batman into your picture with inpainting.
 
This is what "before you were born" generates.

LW1 - 1 (40).jpeg
I agree. I feel this way most of the time.
 
They often have extra arms and legs. Hands are hard, and commonly have 6 fingers or more. Negative prompts help some. You can get perfect hands and poses using controlnet. That lets you extrct poses and expressions from pictures, or make your own.

PS: I added Batman into your picture with inpainting.
I figured, but the original pic really does look like Batman.
 
I figured, but the original pic really does look like Batman.
It was close enough it gave me the idea and made it easy.
 
Walking on Venus :

1000007273.jpg
 
Uncomfortably anatomically incorrect:
1710184764202.jpeg
 
Teenage Mutant Ninja Frogs :

LW1 - 1 (41).jpeg
 
Robot John Travolta :

LW1 - 1 (42).jpeg
 
Back
Top Bottom