I'm so glad bullying worked and csp actually backed down and said y'know what stable diffusion isn't what the folks want and apologised for it with no ifs or buts.

I'm not even an MCU watcher anymore, haven't been for years, but every single person involved with using AI for the Secret Invasion credits can fucking choke. Go eat shit

My Comment on the AI Accountability Policy Request for Comment

I sent in the following comment on federalregister.gov:

---

Hello, thank you for taking comments on this matter.

I have had a strong interest in Artificial Intelligence since I was very young, and I even used some of the first "artificial life" computer software in the late 1990s. I have been studying AI and applications of it for my whole life, and am well-versed in how it works. Currently, I'm pursuing a computer science degree, and my concentration will be in using AI in online security.

That is why I am qualified to say this: Tech CEOs and financers DO NOT understand the technology. I have seen time and time again that the people who use and promote this software don't understand that it stores data from art and writing that it has no right to have. They know how to use it, and they know that it has the potential to generate revenue, but they don't consider the people it hurts. In fact, they don't care about the people it hurts.

Technology should make our lives easier. It should enable us to pursue art, music, and writing at will, in our business AND free time. It should not take those opportunities from us. And if it MUST exist, it must only take from art which it is given. Currently, the art datasets that exist, which are based on LAION-5B and the like, contain literal BILLIONS of "scraped" art pieces, for which the original artists gave no permission. And these works are being used to generate money for those who own or utilize the software, with none of the profit going to the original artists. This is a copyright violation, and unacceptable.

Datasets must only contain Royalty Free/Creative Commons art, AND/OR art which has been commissioned and paid for to be part of the dataset. Should any art exist within a dataset that is NOT RF/CC, the original artist MUST BE COMPENSATED PER USE.

AI must not become a "cheap way to create art," as this functionality would break the backs of artists who are already underpaid. If AI is to exist, it must work IN CONJUCTION with artists to elevate them and pay them fairly. Use of a dataset should pay the ARTISTS, not those hosting the dataset or the software.

If the technology cannot accommodate this, it MUST NOT BE USED until it can.

This is an AI-generated image

This is not a photo. It's not Midjourney output either. Some basement nerd kept cooking and cooking data models for Stable Diffusion until he got to this point. This looks like a hobbyist photographer messing with an entry-level DSLR camera, with mostly correct anatomy and lighting. And what's important, a five year old gaming computer will need just a minute to generate a shot like this.

I could literally take photos like this. Fuck, I did take photos like this. A bit of grit, a bit of fuzz, loose strands of hair sticking out, those pictures are better than perfect - they're believable. Any schmuck can ask for razor-sharp shots of overly photoshopped supermodels with a fuckton of filters on top on Midjourney. Good luck catfishing anyone with those. But this data model, it's so good at being average it's a fucking achievement. You wouldn't look at it twice if I didn't tell you. Those weirdly tall dandelions would go right over your head. You wouldn't even think twice about the nailpolish missing from the thumb but perfect on the index finger. Or that one weird fold on the skirt. AI characters don't have blemished skin after all. Or normal hands.

As a photographer, I'm impressed. As someone who deals with Stable Diffusion, I'm impressed too. And that does mean something.

"The heart is the center of our emotions and thoughts, and it is the primary means of expressing our true selves. It also serves as a bridge connecting our conscious and unconscious minds."

"The heart is the place where we feel and think, acting like a command center of sorts.

It is because of our hearts that our emotions and thoughts are born, leading to our actions.

In other words, the heart is an essential tool for expressing ourselves.

Furthermore, the heart also serves as a bridge between the conscious and the unconscious.

The conscious refers to thoughts and emotions we are clearly aware of, while the unconscious refers to deep-seated thoughts and feelings that, although we are not usually aware of them, influence our actions and emotions.

By connecting these two, the heart allows us to understand ourselves on a deeper level.

Therefore, nurturing the heart is synonymous with nurturing oneself.

By listening to the voice of the heart and understanding our emotions and thoughts, we can confront ourselves and cultivate the strength to live authentically."

I keep making a joke that Ai artists are just the kids that traced in year 7 and told everyone they drew it but now that I think about it some (not all) Ai artists really do have a childish tracers mentality

I mean the ones who won't share their prompts and gloat at artists how their ""art""" is so much better and artists are gonna get replaced. Like the guy who generated a run of the mill one piece image (with some very obvious Ai fuckery in there) when someone pointed it out he told them they could never draw a pic like that

that someone was an animator for one piece.

For a kid who never moves past tracing or Ai artist with that attitude they found the "shortcut" to the end result, they don't care about creation...that's hard they just care about the little buzz they get when they show someone else their pretty picture and say "I made this"

No account sign ups, no tokens to buy or limited uses. Yours for free and just a one click install. You can even use it without a graphics card. Check it out here: LifeHacksThatWork - Ultimate AI Image Generator.

A lot's been on my mind...

A lot of it either feel I can't really talk about or it's personal stuff that I prefer to keep private.

I will say this though: the rate AI is advancing if absolutely terrifying me.

Now, normally I prefer to keep posts as short as possible, so I because this is a rant and kind of spur of the moment (and also a long time coming), so I legit have no idea how long I'm going to run on this

Also want to point out before I get into all this, I don't hate AI, but the ways it's being used.

*TL;DR at the end.

Now, let me clarify, this is not just about the image generators, but everything involving itself in creative fields and possibly beyond. We're basically at the point where you can't tell what's real anymore. I say this as I just prior to me starting to write (March 27th) this I read an article regarding an AI generated pic of the pope that fooled a lot of people.

Then there's the issue that many artist were worried about in regards to losing work to AI. It's already happening, though currently in a different field: modeling. Basically, Levi's is going about testing AI generated models and claiming to do so in the name of 'diversity.'

And probably the biggest issue is people either remain ignorant to what's going on, don't care what's going on, or want AI to take control of everything.

I say of the former two (and basically everyone for that matter), you really should both learn and stay informed as best you can as this is an issue that has an effect on everyone. I also say this as someone who's looked a little further into how some of this tech (image generators specifically) work. Also be aware, just like anything on the internet, there'll be misinformation, so you probably shouldn't take one source as a definitive final answer (This includes myself as I'm just as prone to making mistakes and having bias as anyone else). On top on that, for the love of god, don't get your info from Twitter or Facebook.

Now, as for the latter, I highly doubt it's going to end up being a case of you being free to sit around and play video games all day. If you're hedging on the idea of UBI and live in the US or anyplace where government greed an lobbying is rampant, it's more likely you'll get barely enough to squeeze by, if anything at all.

Now, while none of these problems have easy solutions, there are those that are working for and in favor of people and human generated content. With lawsuits against image and text AI generators and with a group working on software that messes with image outputs I am somewhat hopeful that protections and more ways to fight back will come around.

In the mean time, I'm still going to work my craft, and you should too.

And by that I kind of also mean I'm gonna try to stop holding back on stuff just because I don't have a proper watermark yet. It's another thing that's going to take a while, especially since I'm going to be working with new software for this. In the mean time, I recommend if you do any sort of visual art you do the same thing too. Unfortunately for writers and coders, I don't have any suggestions as that's not my area of expertise (but don't take this as me saying, 'oh, ur on ur own, lol').

Now, admittedly, since I began writing this, some new info has come to light (*insert slowpoke meme here*). Apparently over 1,000 tech leaders, including Elon Musk (if even he is campaigning for this, then something is up), signed a letter on the topic of slowing down or even halting AI development. This is potentially good, but too be fair, it doesn't stop the damage that's already been done.

Like I said above, I'm hoping that protections for people from AI will become a thing.

Sorry if this seems like a downer ending, but I don't really know what else to say, and I'm not really a writer or journalist. This isn't the last I'll be talking about stuff like this though.

TL;DR:

AI is already affecting credibility and people's jobs.

Don't just take something on person of side says as truth; try to find a credible source (i.e., not Twitter or Facebook).

UBI is not an end all solution, especially if your government is known for it's greed.

There are people trying to curb some of ethical the issues from AI.

Keep creating, no matter what.

Tech leaders are campaigning for a slowdown/halting of AI development.

This won't be the last time I bring up stuff like this.

Playing With Ai Image Generation - Episode 2

Welcome back to the second installment of “Playing With AI Image Generation,” where I continue my exploration of the exciting nexus of artificial intelligence and art. This episode once again features Leonardo AI, the remarkable AI platform that’s redefining the contours of digital artistry. Witness as I harness its potent capabilities to create everything from breathtaking digital vistas to…

View On WordPress

AI image generator prompt: The birth of tragedy depicted as a birthday cake as painted by Cy Twombly

Hey... I know I rarely come on here anymore.

I've been playing around with AI image generation with the new ArtBreeder collage feature, and I made a little Dean Winchester short about him getting attacked by a tree monster while he's just walking through the forest haha.

I wanted to see how well this would post. Hopefully, the quality is okay.

I am very torn about the use of AI in “creating” new art.   I agree that simply entering a prompt does not make someone an artist, nor a copyright holder, as these are, by definition, derivative works.  I believe that true artistry requires creative thought.   I argue that AI images are not transformative, as something Warhol might’ve been.  That’s a fine line, and subjective; “transformative” is nebulous (and, well, “art” is most certainly a fluid term, as is “originality”—when combining prompts to create something of the likes of which has never been seen before, is that original?).  Where does Marcel Duchamp fit; does the associated statement/thought behind the artwork make the piece transformative?  Does an artist have to be able to explain the artistic merit/justification of their work, or should it be able to stand on its own?  Does the interpreter have any say in the matter?   I do believe that it can be useful as a guide for other things.  I’ve got many ideas for a book—a title, characters, a storyline with a healthy lesson for life, some scenes in my head, but haven’t started putting pen to paper for those thoughts (and I won’t share them here for fear of someone stealing my ideas).  I’m thirsty for seeing it come to life, however, despite my inability to write (I would consider myself a non-fiction/academic writer, not a creative one).  So, using an image creator like the one released by Microsoft today (powered by DALL-E; or Stable Diffusion, or whatever), I can put together a storyboard where words can be filled in later, perhaps with a co-author (and original illustrator).  The above image roughly matches part of a scene from the story; I asked the image generator to use the style of Studio Ghibli’s ‘Kiki’s Delivery Service,’ though I disagree it did a good job on that element specifically (the results when I ask it to be styled like a Renaissance painting are incredibly similar to this, so something’s off there).  Though not originality, AI can provide assistance.   I’d love to have some sort of philosophical conversation about the whole thing, to better understand my own ideas, and those of others.  I’ve got a number of mutuals that are artists.  This is still an area in its infancy, and so are opinions on the matter.  If someone wants to send an ask, that would be welcome.  

Everyone who supports AI art, I guess is okay with consuming for the sake of consuming and probably is totally cool with all TV, games, books, art, music etc. being generated by a few words typed into a generator in the next few years I guess

Welcome to the third installment of "Artists Through The AIs": this time, we're going way back to the classics with a certain young witch dressed up as a certain sorceress.

Due to odd behavior of the data model (the freshly updated Dreamshaper V7), I had to fiddle with the prompt. Namely, if emphasis was put on the "painting" part, the images were sprouting easels and/or heavy gilded frames every-goddamn-where, even with CLIP skip set at 2 instead of the usual 1. So "painting" was left at 1.0 weight, while "by X" was emphasised to 1.2. Other than that, the prompt is "embedding:kieshi3v with long (red hair):1.2, wearing long red medieval dress with corset, medieval castle exterior background, dramatic lighting, detailed, high definition", and the seed is 254762234227837. Negative prompt includes "((easel)), ((frame)), ((interior)), red eyes" in addition to my usual heavily magical list. So here we go.

Leonardo da Vinci - I'm having second thoughts about the background common for several of the renders, but there's enough distinction between them to highlight Leonardo's style. The main focus is skin shading and curly hair, as seen in Leonardo's depictions of Virgin Mary.

Caravaggio - deceptively similar to the previous one, but with a noticeable distinction: the deeper shadows and higher contrast, creating a very Caravaggian chiaroscuro.

Titian - asking for "Titian" and "Tiziano" generated very similar takes in a lamp-lit corridor with way too modern look and pretty much no similarity to the paintings you can view online.

It also appears that AI omits the paint and canvas textures from Renaissance paintings, something it didn't do with the impressionists previously. This is most probably due to less emphasis on the "painting" part, as I'm going to show you in another post. Moving most of the description two-three tokens down apparently also creates a new baseline with a window or portal (either fitted with a door or not) instead of an overcast outside shot.

Albrecht Durer - whether you spell him with an umlaut or not, the outcome is a generic take with a portal in the background. The umlaut just removes the door and gives Sabrina a handbag.

Lucas Cranach the Younger - what we have here is a miscommunication. If you ask for the art of Lucas Cranach, sure as shit you'll get your character in a black beret popularized by the elder one's paintings of German VIPs. However, the diametrically different interior characteristic of Renaissance painting and less pronounced effect of the embedded likeness of Kiernan Shipka makes me think that, once again, we're dealing with a "So you have heard of me" behavior.

Hans Holbein the Younger - once again, something's way off here. We have a similar Renaissance interior background with some traces of brushwork, less pronounced embedding and some odd lace bonnet, but nothing that would link the work to either of the two Holbeins. I think that we're dealing with the side effect of the prompt being knocked two further tokens down, but I'm not sure. It goes too far, particularly if you consider that differences between other images are much less pronounced.

Rembrandt - as in the case of Caravaggio, the distinct part of this one is the chiaroscuro. There's also more pronounced detailing of the collar and shoulders of the dress, possibly inspired by the outfits in "The Anatomy Lesson of dr Tulp", "Syndics of the Drapers' Guild" and "The Night Watch".

Vermeer - there's no doubt. This is Vermeer. Two different seeds and I got two very similar mashups of "The Girl Reading A Letter", "The Girl With Pearl Earring" and "The Milkmaid". This goes so far against the prompt that we're dealing with a similar case to Simon Stalenhag from the first test.

Peter Paul Rubens - this one itches my brain something fierce. While scrolling through the renders, this one and four neighboring ones share the exact same position of white shift and a dark underbust corset, meaning they're merely slight variations on a baseline, but I swear that I've seen a similar painting of a long-haired man in a black doublet with frilly white collar and sleeves. Was that a Rubens? I have no idea (it wasn't Hals' "The Laughing Cavalier", though).

Diego Velazquez - this one is trying to do something. Once again, we have a portal in the background, but the character's position is slightly off the baseline and the palette is slightly muted.

Murillo - way too generic. The background is curiously off, as it's not a portal but a vaulted ceiling supported by pillars, but the pose is shared by four other renders, including the Rubens one, meaning it's sticking to the baseline fairly closely.

Francisco Goya - this is an odd one. It has nothing to do with Goya's works, but still, the pose is off the baseline, the background features brushwork and an abrupt switch from a near wall to a wider expanse with bizarre levitating architecture. I have no idea what's going on with it.

Eugene Delacroix - it's generic. Sure, the background features an open street with some people as opposed to the typical portal and lamp or chandelier, but the pose and outfit are kinda samey. What, no attempt to go anywhere near "Liberty Leading The People"?!

Jacques-Louis David - another generic one. I mean, what the fuck. If you look at David's corpus of works on Wikipedia, this could have been done entirely differently, yet we have an underbust corset, a portal and a light source in the background.

Thomas Gainsborough - finally something. Discernible brushwork, unusual contrast and lighting, and - as a bonus - the Strawberry Hill Gothic castle in the background make this one distinctive and consistent without plowing straight into obnoxious homage slash blatant plagiarism territory.

I dare you to look up your favorite Pokemon on one of those AI image generators

Not technically my absolute favorite, but I did try looking up ninetales recently and that’s def one of my favorites

@staff

OUR CONTENT SHOULD BE OPTED OUT OF AI TRAINING BY DEFAULT!