Jump to content
IGNORED

AI Art


Squee

Recommended Posts

3 hours ago, Wunderbar said:

why does it struggle with faces so much compared to other parts of a human?

It's mostly because mistakes in a face are more noticeable than mistakes elsewhere. The creator of Craiyon (formerly known as Dall-E mini) explains it here: https://www.sciencefocus.com/news/dall-e-mini-creator-explains-blurred-faces-going-viral-and-the-future-of-the-project/

Quote

“The image is encoded into a very shot sequence of numbers so that the model can learn faster. Because of this, the model makes a lot of mistakes. However, when you draw the Moon, a landscape or a tree, you don’t really notice the issues there.
“When it is on a face, we pay a lot more attention. If the eyes are out of order or the nose is misshaped, it is weird. It is the same on animals and cartoon characters, it’s just something we pay more attention to than misshaped objects. Really, the model is equally good or bad at everything.”

The same goes for AI upscaling algorithms too. There are general purpose AI upscaling algorithms that can upscale any kind of content: landscapes, buildings, etc. To human perception, the faces upscaled with these algorithms look horrible most of the time. The only AI upscalers that do well with faces have specialised algorithms and training data that exclusively deals with faces.
 

  • Like 2
Link to comment
Share on other sites

18 hours ago, Wunderbar said:

why does it struggle with faces so much compared to other parts of a human?

If you ask for faces or portraits specifically, it does better, and you can ask for certain facial details to increase your chances. Ideally you'd reroll until it gets the right result. Back in my painting circles we had the saying "a bad face ruins everything.", meaning that you'll get away with a lot of wonky stuff except for the face. Aside from what ghsotword brought up, there's a lot of preventions built into the AI to avoid the likenesses of real people etc. I think that contributes to messing it up sometimes. 

722418899_DALLE2022-08-0519_08.48-AphexTwininasecretspymeetup50sfilmnoirsmokingacigardetailedphoto.png

  • Like 1
Link to comment
Share on other sites

1 hour ago, chim said:

If you ask for faces or portraits specifically, it does better, and you can ask for certain facial details to increase your chances. Ideally you'd reroll until it gets the right result. Back in my painting circles we had the saying "a bad face ruins everything.", meaning that you'll get away with a lot of wonky stuff except for the face. Aside from what ghsotword brought up, there's a lot of preventions built into the AI to avoid the likenesses of real people etc. I think that contributes to messing it up sometimes. 

722418899_DALLE2022-08-0519_08.48-AphexTwininasecretspymeetup50sfilmnoirsmokingacigardetailedphoto.png

You still paint ? How does this ai stuff compare to that for you ? you get the same kind of enjoyment playing with that kind of stuff?

Link to comment
Share on other sites

1 hour ago, Wunderbar said:

You still paint ? How does this ai stuff compare to that for you ? you get the same kind of enjoyment playing with that kind of stuff?

Not super often these days. I don't really look at them as similar in any way, even though I mainly painted digitally. This is more of a novel dice roller gadget with visual results. So it's pretty cool but it's nowhere near the enjoyment, effort and involvement of constructing something stroke by stroke. There are lots of aids and tricks in the digital painting world and I always disliked relying on those, but I'm also aware there's a world of traditional painters who think all digital art is BS.

You could tie paintbrushes to your dog's paws and claim you're the creator of whatever happens. It's easier to feel removed from what Dall-E does as it's pretty unexpected. I'm not sure how I'd feel about a version with super specific instructions or direct brain link. It's a bit like generative music versus playing an instrument, I'm not saying one is strictly better but they are different. 

 

 

  • Like 1
Link to comment
Share on other sites

7 hours ago, chim said:

Not super often these days. I don't really look at them as similar in any way, even though I mainly painted digitally. This is more of a novel dice roller gadget with visual results. So it's pretty cool but it's nowhere near the enjoyment, effort and involvement of constructing something stroke by stroke. There are lots of aids and tricks in the digital painting world and I always disliked relying on those, but I'm also aware there's a world of traditional painters who think all digital art is BS.

You could tie paintbrushes to your dog's paws and claim you're the creator of whatever happens. It's easier to feel removed from what Dall-E does as it's pretty unexpected. I'm not sure how I'd feel about a version with super specific instructions or direct brain link. It's a bit like generative music versus playing an instrument, I'm not saying one is strictly better but they are different. 

 

 

Good point. I think it would be better to think of anything you do as yours on some level. Even if you only gave it a prompt, you gave it existence. I think we should think of Ai Art as an extension of ourselves in an abstract sense this way. Especially since I kinda understand neural networks intuitively now, it just makes sense to me. Your brain does the same think the ai does in this art. It builds in the background out of small fractal variations, presents it to other neural networks designed to judge, and your brain sifts thru it with more neural networks. Always commenting on itself and delivering meaning to this or that.

I'll admit it's a romantic way of looking at it but that's judging from feeling for you.

Edited by Brisbot
  • Like 1
Link to comment
Share on other sites

A cyberpunk goat riding a Harley Davidson

 

DALL·E 2022-08-08 21.54.09 - A photo of a menacing cyberpunk goat riding a Harley Davidson in the night, detailed neon 80's.png

DALL·E 2022-08-08 21.50.21 - A photo of a menacing cyberpunk goat riding a Harley Davidson in the night, detailed neon 80's.png

DALL·E 2022-08-08 21.50.20 - A photo of a menacing cyberpunk goat riding a Harley Davidson in the night, detailed neon 80's.png

DALL·E 2022-08-08 21.50.17 - A photo of a menacing cyberpunk goat riding a Harley Davidson in the night, detailed neon 80's.png

A labcoat scientist experimenting with a monstrous modular synthesizer 

 

DALL·E 2022-08-08 21.55.02 - A 50's black and white photo of a scientist in labcoat experimenting with a monstrous modular synthesizer.png

DALL·E 2022-08-08 21.55.00 - A 50's black and white photo of a scientist in labcoat experimenting with a monstrous modular synthesizer.png

DALL·E 2022-08-08 21.54.58 - A 50's black and white photo of a scientist in labcoat experimenting with a monstrous modular synthesizer.png

DALL·E 2022-08-08 21.54.54 - A 50's black and white photo of a scientist in labcoat experimenting with a monstrous modular synthesizer.png

DALL·E 2022-08-08 21.59.09.png

DALL·E 2022-08-08 21.59.03.png

DALL·E 2022-08-08 21.59.00.png

Edited by chim
  • Like 1
Link to comment
Share on other sites

  • 5 weeks later...

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.