AI generated images

[RedOrc]

It is clear to me that using an algorithm to generate images is not a particularly creative or aesthetic process. But I'm still getting away from the fascination of what computer technology takes us to imagine. I can neither look at it, nor look away.

The first picture has a really nice face, while most of the rest are deeply into the Uncanny Valley. I am disturbed how "good" many of the pictures are. I would not thave thought artist was a job at risk of being replaced by computer.

 

[Davec5299]

I guess a lot of artists agree with @montycrusto on this.....

https://www.bbc.co.uk/news/technology-62788725

 

[montycrusto]

I guess a lot of artists agree with @montycrusto on this.....

https://www.bbc.co.uk/news/technology-62788725

Yeah it’s absolutely gross, especially with that hit-list of all the artists whose work you can rip off using this software. And I’m also somehow offended that I’m not on the list..

 

[fnhsr]

I wouldn't be so pessimistic as to say that AI-generated images herald the end of handmade art. If you look at the results, you can see a certain monotony. There is simply a lack of personality. And these images really don't go that harmoniously. But the fascination remains with me. That's why there are more AI generated images.

 

[fnhsr]

And more.

 

[fnhsr]

Never ending.
Did I ever mention that in real life I make my living doing art and creative crafts, and that quite successfully? This AI frippery here is not meant to snub artist colleagues...

 

[montycrusto]

But the fascination remains with me. That's why there are more AI generated images.

I hope it’s only a passing phase.. you could easily generate thousands or millions of images this way; I’m not sure CF has room for such a banquet..

 

[mp5stab]

Can't tell if this thread is frustrating or inspiring. Fact is, this might just be what photography was to painting. I always use whatever tools I have available (I went from pencils, to inks, to digital art, to 3d art; I have no loyalty to one production method) so I think anything to get me the image I want is a good tool. But this changes a lot of things in fundamental ways. I think AI generated art could massively increase the availability and ubiquity of art in most spaces, but also drastically devalue it. I think it may spurn on a number of creative new ideas, but also, devour itself in a cascade of self-referential psuedo-plagiarism.

It's also telling that these models have been trained on classical paintings in the public domain. They have the quiet stoicism of martyrs, and aren't screaming bloody agonized messes as we usually depict them. No doubt you could whip something like that up as well, but I suspect the power of these models may be limited by their training data. Once you can train a model on say, the collected works of jucundus, skatingjesus, gerd1000 et al, and bias the style results accordingly, then I suspect you will have a series of images not much removed from the type of thing the forum wants to see!

I also wonder how good these models can be about specifics. Like, if you asked for every woman to have her hair in a bun or ponytail, instead of down loose, would it manage? The truth is, no doubt, that this kind of development is only a few years away regardless.

 

[montycrusto]

I just came across this account on DA

https://www.deviantart.com/saidbgz

I think it’s all done with this kind of software, parasitising the work of real artists, what do you guys think?

 

[bobinder]

I just came across this account on DA
https://www.deviantart.com/saidbgz I think it’s all done with this kind of software, parasitising the work of real artists, what do you guys think?

One humanoid figure has three arms, and another has three legs - I suspect you may be right, Monty.
And he says, 'Make sure to follow - there is a lot coming soon'...

 

[Barbaria1]

Is there a concern that the scale and regularity of the output, unless controlled or nuanced, could be simply overwhelming? So many variations but also a pleasure-deadening familiarity/redundancy to it all? I know little about this. So don’t take me too seriously, but one should always ask the question in a world where innovation and AI seem too be on the march everywhere.

 

[RedOrc]

Many of the pictures have this strange thing with the top of the head out of frame. I wonder why it does that?

This was pointed out to me by a friend who was playing around with it on his phone, so it is not just the crux pictures.

 

[montycrusto]

Many of the pictures have this strange thing with the top of the head out of frame. I wonder why it does that?

This was pointed out to me by a friend who was playing around with it on his phone, so it is not just the crux pictures.

Yes, the cropping is often extremely odd; that’s one of the signs to look out for when spotting this stuff!

 

[Eulalia]

The fact it can be done doesn't mean you have to do it and post every single image here. I'm sure members here would be interested to see a selection of what AI users think are the very best outcomes, just a small selection at a time, with some discussion of why they consider them exceptionally successful - don't just swamp the site, please.

 

[The Fallen Angel]

It's interesting to see how these A.I. pics have rapidly developed from "Aliens Crucified" by Salvador Dali to Dr. Frankenstein's Females Freak Show. Please supply more info on how people can make these then perhaps we could see the best of what can be done.

 

[Jollyrei]

Is there a concern that the scale and regularity of the output, unless controlled or nuanced, could be simply overwhelming? So many variations but also a pleasure-deadening familiarity/redundancy to it all?

That was my thought as well. A promising sort of method for producing images, with the "art" in finessing the algorithm for the AI images. Some of these are quite good, but there is a certain "sameness" to them which potentially "devalues" the best images. A curated set of postings of the best works would be a suggestion.

 

[montycrusto]

Saw this the other day and thought of this thread.. a moving story of a man’s love for a cabbage…

 

[malins]

It's also telling that these models have been trained on classical paintings in the public domain. They have the quiet stoicism of martyrs

also the AI is really into flowing loincloths and drapery...

Again though we can see that these systems have no internal representation of what they produce.
For instance even a beginning artist won't sometimes get confused about how many belly buttons a human body is supposed to have...everybody's internal model of how a human body is built requires exactly one, and you won't get two or three in some percent of cases.
(While of course an advanced artist's internal model of the human body will include much more anatomy than the general person)

Also, the AI has picked up the notion of a crown of thorns or a halo around the martyr's head but seems to often merge it with the hair.

Again it doesn't have a model of what a halo or thorn-crown is as an object and a symbol, the data just contains lots of examples with 'ring-like structure close to / around / above the top of the head'. So with a certain probability it tries something like that.
Things melting into each other is a bit of a general problem.
All of this of course can and will be fixed by filtering the generated images but even if these obvious machine-typical errors aren't apparent anymore it is still a work of 'blindsight'...

I also wonder how good these models can be about specifics. Like, if you asked for every woman to have her hair in a bun or ponytail, instead of down loose, would it manage? The truth is, no doubt, that this kind of development is only a few years away regardless.

I guess so long as there are enough images of the hairstyles categorized with the right names in the dataset it will work very well.
However someonoe could describe to you, without knowing its name, a certain foreign/historic style you've never drawn before, and you could work out roughly what the result should look like. The AI model just has data about different looks associated with the terms but can't 'build' one.

As for the impact on art as such ... I guess the impact will be very rapid for mass-produced imagery that no one really looks at too closely. So the random shiny happy people that are portrayed in the background of products/ads will soon be all AI-made. Same for random stock photos of sharply dressed businesspeople for website banners and so forth. All art-hosting websites will have to deal with the issue of getting flooded with such imagery.

 

[Jollyrei]

everybody's internal model of how a human body is built requires exactly one, and you won't get two or three in some percent of cases.

Unless you're Picasso.

 

[malins]

Unless you're Picasso.

Oh I think Pablo's internal model of the human body was alright, he was perfectly able to do naturalistic drawings,


just chose to consciously transgress them. And well, love him or hate him there was a method to his transgressions