AI generated art

[THX1138]

Never tried it, but I saved it. When I have time, I'll give it a go!

 

[Foxbat]

Trying some stellar nebulae

 

[Foxbat]

Still at it...

 

[Cthulhu.Science]

Exploring the world of Simon Stålenhag

 

[Cthulhu.Science]

 

[Cthulhu.Science]

After writing a prompt -
I usually choose 3 images at a time. It seems like an efficient way to see if I'll get good results.

Also, spinning through each of the "Art Styles" can be a fun and interesting exercise.
For any given prompt, the results can be surprising.
Most of the "Art Styles" will give relatively uninteresting results for any given prompt -- But often there is an unexpected great result from an "Art Style" that you'd never look at. 50s Enamel Sign??? MTG Card??? Flat Style Logo???, not that I've ever had GREAT results from any of those specific styles -- wait, I think I might have.

 

[Foxbat]

View attachment 128018

This makes me think of what would happen when Dune meets Pacific Rim

 

[Cthulhu.Science]

This makes me think of what would happen when Dune meets Pacific Rim

As drawn by Simon Stålenhag. Known for Tales from the Loop -- below is not AI - this is from one of his books.

 

[Lumens]

Simon still outshines any AI produced art. I grew up in Sweden and he captures that atmosphere elegantly.

I think you can still produce good pictures with AI but it takes a lot of work and some understanding of compositon, plus technical knowledge of the medium you aim to emulate, like camera aperture or use of light and shadow etc etc. And probably much much more.

 

[Cthulhu.Science]

Simon still outshines any AI produced art. I grew up in Sweden and he captures that atmosphere elegantly.

I think you can still produce good pictures with AI but it takes a lot of work and some understanding of compositon, plus technical knowledge of the medium you aim to emulate, like camera aperture or use of light and shadow etc etc. And probably much much more.

I think of Text-to-Image generation of images similarly to the way I think of cameras. It is a tool to create images.

In the 19th century some artists worried that photography would be the end of painting, and their careers. The death of art as been greatly exaggerated.

AI can attempt to create images in the "style" of various artists -- but artists have been generating art in the "style" of other artists for as long as there have been art. Not to mention actual forgeries.

If we compare planet Earth of 2024 with planet Earth of 1824 are more pictures produced by brush today or fewer. I do not know. My gut instinct is that more traditional (by brush) paintings are produced today by an exponential factor. There are lots of reasons why. First is that each artist does not have to prepare their own canvass and collect the materials to produce their own paint. Or have a staff to do it for them. Access to materials is generally ubiquitous and cheap. So cheap that school children are given materials and invited to produce vast quantities of brush on paper "works of art," as millions of refrigerators can attest to. In addition, many more adults today have the time available to them to pursue the endeavor.

The point is that photography and MSPaint and photoshop and Text-to-Image AI are not the demise of art. AI is simply a tool to create images. In the process, like the many new tools before it, Text-to-Image generation provides access to the creation of images to people who would not create images otherwise.

 

[Cthulhu.Science]

Simon still outshines any AI produced art. I grew up in Sweden and he captures that atmosphere elegantly.

I think you can still produce good pictures with AI but it takes a lot of work and some understanding of compositon, plus technical knowledge of the medium you aim to emulate, like camera aperture or use of light and shadow etc etc. And probably much much more.

Netflix did a series "Tales from the Loop." It was weird, as one would hope. The stage dressing, props, setting were right on target.

The story was appropriately weird but missed the mark somehow, which is too bad. I'd love to see his world depicted over and over and over again. I'd love to visit Stålenhagland.

I went through a lot of AI prompt wordsmithing to get something on target, but I was never able to get the depth of Stålenhag. In the image above there is so much more going on than the big red "thing" in the middle of the image. We can ask, "What is that?" is it an automaton? is it a vehicle? But the rest of the image is so detailed as well. Is that a floating traffic signal? what color are those lights and why are they the same? What are those kids doing and wearing? lets look at the two buildings in more detail and all the clutter everywhere. Stålenhag is amazing in the details, something AI simply does not capture.

 

[HareBrain]

Netflix did a series "Tales from the Loop."

I wish it was on Netflix, but it's Amazon Prime.

I used to follow Stålenhag on Twitter, and he was not at all happy with the idea of generative AI being trained on his work.

 

[Lumens]

I think of Text-to-Image generation of images similarly to the way I think of cameras. It is a tool to create images.

Yes exacly. Don't let me put you off from using it, I just think it requires a very different skill than for example painting or photography, even though you have to base your prompts or image inputs on those skills to get the best out if it.

I can understand professional artists having an issue with people putting "In the style of..." prompts in their browser to emulate a skill they themselves have spent a lifetime perfecting. In fact it must be heart breaking. I personally don't have strong opinion about it one way or another. One thing is for sure, it's an interesting future we live in.

 

[Cthulhu.Science]

 

[Elentarri]

The tiger part of your creature looks a bit miffed!

 

[Cthulhu.Science]

Same prompt as above -- different "style"

 

[Foxbat]

After spending a bit off time playing around with it, I have to say that it's fun but (there's always a but) I've discovered that there's one massive difference between this and other forms of art.

I still haven't got much of a clue what I'm doing and I'm just playing around with words. I have no real direction or purpose and yet I can produce some fairly decent results. If I were to take the same haphazard approach to a canvas and oils, I'd end up with the results more akin to a three year old splashing paint everywhere (and getting most of it on myself). The results of such an experiment could be even worse than that, they might be mistaken for Modern Art

This is as close to Geiger's xenomorph I've managed to get so far.

 

[HareBrain]

Netflix did a series "Tales from the Loop."

I wish it was on Netflix, but it's Amazon Prime.

Looks like the Christmas fairy was listening, kinda.

Oddly enough I've just seen advance notice on Netflix of a film called "Electric State" which is based on Stålenhag's third book (and the one with the strongest narrative). It's clear, though, that the film goes way beyond the limits of the book's story, and will probably therefore be utter pants.

 

[Foxbat]

Trying some looney tunes types

I've found that the AI sometimes has trouble with living forms (see the duck bill in first picture). I'm not sure if that's a flaw or if I need to be more precise in my wording.

A couple of comic covers.

 

[Toby Frost]

Leaving aside my many issues with AI for a moment - what the heck is she wielding?!