What art have you been generating?

I leave it as an open question, but my point in posting this to the thread is to showcase my early experiments with a new art-generating A.I.

I'd been having great results using Wombo to generate art, but this new generator in beta was producing such stunning results that I decided to knuckle-under and subscribe so I could use it to generate art for my RuneQuest: Glorantha TTRPG campaign. Here's one of the Red Moon Goddess:



Another of a tribal Orlanthi people overseen by their deified ancestors:



And even this re-imagining of my very own Avatar that was originally from Bakshi's "Wizards":




So far I'm **very** pleased with the results. If anyone else wants to dork around with it, the name of the generator is Midjourney and I use it within Discord.

I have tried that before, and they have turned out well, but there is a new one, called Stable Diffusion, that my friends have been using to great effect. I cannot use it because of my video card (hoping to change that in the somewhat near future). It is free, and here is the installation guide: rentry.org/voldy#-guide-

Here are some examples: mpost.io/best-100-stable-diffusion-promp...xt-to-image-prompts/

Whoshim wrote: I have tried that before, and they have turned out well, but there is a new one, called Stable Diffusion, that my friends have been using to great effect. I cannot use it because of my video card (hoping to change that in the somewhat near future). It is free, and here is the installation guide: rentry.org/voldy#-guide-

Here are some examples: mpost.io/best-100-stable-diffusion-promp...xt-to-image-prompts/

That is a great tip! I'll have to look at this one as well. I didn't realize the computation was local to the vid card. I have an NVIDIA 3070, which is pretty capable. I'll give it a whirl.

So if you put the same text into the engine will it converge to the same image, or do you get a different one every time?

Those images look very nice. I take issue with calling it "art". I don't believe there is such a thing as AI Art; there is only AI image generation. Art requires a human and is a personal endeavor that a computer will never understand, at least in our lifetimes.

I had fun messing around with it a few months ago. Just did some basic stuff and didn't really bother trying to refine them.







A couple of Patreons I subscribe to for 3D printing models have been using it to generate concept art for models that look really cool, especially when you take the time to iterate on a prompt:

I feel like I've seen SO MANY movies/TV shows lately that were written by a script AI version of this technology.

Plot feels vaguely familiar or cobbled together from other things, scenes that are kinda disconnected from each other in an oddly disjointed way, dialogue that feels like two people talking past each other without a deep understanding of what they are trying to say.

I give it 15 years before I can generate an entire anime saga from this stuff.

"Hey google, create a 2 hour OVA of robotech meets Urotsukidoji in the style of Hayao Miyazaki........AHHHH transforming tentacle mecha, but they look so cute!"

hotseatgames wrote: Those images look very nice. I take issue with calling it "art". I don't believe there is such a thing as AI Art; there is only AI image generation. Art requires a human and is a personal endeavor that a computer will never understand, at least in our lifetimes.

I was quick to agree with this, but gave it a second thought. There are humans at every corner with this. The AI is assimilating human thought and artifice and transforming it, which, I believe, is what a cortex does. Isn't this perhaps AI augmented art, but art just the same?

Sagrilarus wrote: So if you put the same text into the engine will it converge to the same image, or do you get a different one every time?

TBH, it's a complex question to answer, partly because I'm new to it. I would say "different each time", but sometimes it will seem to take a basic arrangement and generate something kind of similar, but not always. Also, you can request it to take images it has generated and create new variations on that image. Also, some of the features can narrow your results it seems, such as the --hd feature. But I've gotten striking results with the realism of the --hd option to generate images in the style of known photographers, such as JP Waitkin. I put something like "antique biology lab in the style of JP Waitkin into the generator and got this:



Asking it to put variatons on single image generated, it produced similar outcomes. This is as narrow as I've seen it generate:

Amontillado wrote:

hotseatgames wrote: Those images look very nice. I take issue with calling it "art". I don't believe there is such a thing as AI Art; there is only AI image generation. Art requires a human and is a personal endeavor that a computer will never understand, at least in our lifetimes.

I was quick to agree with this, but gave it a second thought. There are humans at every corner with this. The AI is assimilating human thought and artifice and transforming it, which, I believe, is what a cortex does. Isn't this perhaps AI augmented art, but art just the same?

Obviously the question of "what is art?" has always been up for debate. I don't think art applies here because it's just someone typing in a prompt. If I commission a painting, it doesn't make me an artist.

hotseatgames wrote:

Amontillado wrote:

hotseatgames wrote: Those images look very nice. I take issue with calling it "art". I don't believe there is such a thing as AI Art; there is only AI image generation. Art requires a human and is a personal endeavor that a computer will never understand, at least in our lifetimes.

I was quick to agree with this, but gave it a second thought. There are humans at every corner with this. The AI is assimilating human thought and artifice and transforming it, which, I believe, is what a cortex does. Isn't this perhaps AI augmented art, but art just the same?

Obviously the question of "what is art?" has always been up for debate. I don't think art applies here because it's just someone typing in a prompt. If I commission a painting, it doesn't make me an artist.

Not trying to be too heady about it, but while I agree that commissioning a painting doesn't make you an artist, resulting art does imply an artist. I think something like the Turing Test applies here. If I show you a selection of art and you can't differentiate human ones from the a.i. ones, are you holding out that there is some "real" art in the selection, even though you can't tell which it is?

hotseatgames wrote: Obviously the question of "what is art?" has always been up for debate. I don't think art applies here because it's just someone typing in a prompt. If I commission a painting, it doesn't make me an artist.

From my understanding, something that can be done in Stable Diffusion is to mask certain parts of the image that you wish to remain unchanged, and then generate other parts of the image again with a different prompt. So this allows the human to be closely involved in the final product.

I don't intend to get into the debate about what constitutes art*, but I was wondering if the above information may change your opinion at all.

*However, I will state my view here for context. I generally hold to the idea that anything created with aesthetics in mind constitutes art, so almost everything manmade is art to one degree or another. This lines up nicely with the older definitions of art: skill/knowledge (see also the German "Kunst"). I feel that a human is using a tool (AI) to generate something aesthetically desired, and so it is art.

That's interesting. If the human "assembles" the final image but the AI is doing all the actual drawing/rendering, is that really any different than using CAD and computer based animation software? At what point are the tools and the artist distinct? Replicate the artists exact strokes but replace the oil brush with water color, a pencil, or even a finger and the end result would be vastly different, so how much credit goes to the tool?

I'd argue that "modern" art, so reliant on audience perspective, is at HIGH RISK for being replaced by AI. Compare it to the Flemish masters, for example, with their ultra-precise paintings loaded with in image symbology (the right hand holds an apple which means XXX type of thing) and I don't think AI constructs would be much of a threat, hell, even photography struggles to realize all the detail those guys do.

jason10mm wrote: I'd argue that "modern" art, so reliant on audience perspective, is at HIGH RISK for being replaced by AI. Compare it to the Flemish masters, for example, with their ultra-precise paintings loaded with in image symbology (the right hand holds an apple which means XXX type of thing) and I don't think AI constructs would be much of a threat, hell, even photography struggles to realize all the detail those guys do.

Do you have a specific artist/work in mind for that criteria?

"You Yankees are so silly about matters of the heart. Don't you know that women are the only works of art?"

-- Don Henley