Tag Archives: artificial intellegence

AI-generated images BANNED!

Confronted with an overwhelming amount of artificial-intelligence-generated artwork flooding in, some online art communities have taken dramatic steps to ban or curb its presence on their sites, including Newgrounds, Inkblot Art, and Fur Affinity, according to Andy Baio of Waxy.org.

Baio, who has been following AI art ethics closely on his blog, first noticed the bans and reported about them on Friday. So far, major art communities DeviantArt and ArtStation have not made any AI-related policy changes, but some vocal artists on social media have complained about how much AI art they regularly see on those platforms as well.

Getty images has also banned AI generated artwork from their site. From The Verge:

Getty Images has banned the upload and sale of illustrations generated using AI art tools like DALL-E, Midjourney, and Stable Diffusion. It’s the latest and largest user-generated content platform to introduce such a ban, following similar decisions by sites including Newgrounds, PurplePort, and FurAffinity.

Getty Images CEO Craig Peters told The Verge that the ban was prompted by concerns about the legality of AI-generated content and a desire to protect the site’s customers.

I’m not sure banning AI Art is possible. Right now, for myself, I’m experimenting just to see what this new technology will come up with. If the rendered output is less than perfect I just leave it as-is. It’s a technology artifact meant to document the state of the art. As a result it’s easy to spot one of my AI Art pieces especially if a character in the picture has six fingers or creepy eyes.

Other artists continue perfecting their work in Photoshop etc. Depending on their process the final piece may only be partially AI generated. How are these bans going to work? How can a site tell which tools you’ve used to create a digital piece? The genie is out of the bottle. We can’t go back.

Today I let Artificial Intelligence draw what it wants

Leave a reply

I’m continuing to check out art created by artificial intelligence at craiyon.com. You type in some words and the AI creates nine pictures from the prompt.

Yesterday I was trying to see if the AI could capture emotion on a face. It took some iterations but the answer? YES. A big yes. The final few renders were amazing. Here’s an example that I didn’t post yesterday:

the face of a cyborg overwhelmed with love in a painterly style

To get a good result the AI needed lots of very specific text telling it what to do and lots of trial and error to find the right words. The AI needed to understand the meaning and context of the words correctly. When it didn’t things fell apart and became superficial and symbolic, like bland clip-art.

But what if I started from scratch and gave the AI as little direction as possible? Where would it go? Will anything dramatic come out?

Lets begin with one word:

face

OK, not bad. Since I’m giving as little direction as possible, I’ll do two renders of every prompt just to see more variation.

There’s some drawing style in this second one but essentially the same theme.

a face

Things are already getting super cool and abstract. Let’s keep going…

the face

Changing the article seems to change the style. There’s an interesting lack of color in most of the pictures. I also get an old silent motion picture vibe… maybe even some German Expressionism.

These are all wonderfully artistic with deep expression and dramatic lighting. They are way, way beyond the bland images I was expecting.

my face

That changed the style quite a bit probably because the craiyon AI is now referencing all the selfie avatars of the world. That large pictorial reference contaminates this minimalist experiment a bit I think. And what’s up with all the messed up noses? Anyway, continuing on…

your face

The thing I find interesting here is that “my face” really injected a sense of photo realism because of the selfie avatars but “your face” leans more toward cartoon drawings.

her face

Not sure what to think here. The middle bottom of the first render is really creepy with what looks like a two fingered hand.

his face

This is getting very strange. Do I see monkeys in some of these?

its face

Definitely seeing monkeys again.

our face

These are wonderful. Boy, if you ever need a creepy avatar for something, craiyon is the place to go.

no face

Uh oh… There’s a relatively well known character called “no face” in the animated film Spirited Away. Craiyon has choked and reverted to referencing character images as well as using symbols for concepts like the red circle with a line for “no.” It’s also bringing in a t-shirt again presumably because somewhere the no face character is for sale on t-shirts. This is definitely a fail for the AI. I should go back to something that worked, but first I absolutely must try the opposite of this prompt…

yes face

Oh no! Emojis! The ultimate symbol for emotions. Super FAIL! OK fun’s over. Time to get back to the good stuff. Lets go back to something that looked artistic and dramatic…

a face

I see people specifying different “Iterations” on Craiyon. Not sure what it does but I’m going to try it.

a face, iteration x

Hmmm… Flat abstract line drawings. Let’s try a few more…

I’m not sure why all these are line drawings now. I’ll have to read up on what “iterations” do. Maybe I should try something more descriptive to add, like trying to make it a pencil sketch.

a face, sketchy

Not bad. The single word “sketchy” could be taken several ways, but it understood what I meant. Not a bad sketch top middle either, for an AI that usually makes horrible faces. What other description can we try?

a face, shadowy

WOW!!! I need to try a few more of these!

Amazing! Dramatic! Haunting! I could play with this all day!

Check out Craiyon for yourself at craiyon.com.

Can Artificial Intelligence Draw Emotion in a Face?

Leave a reply

I’m continuing to check out art created by artificial intelligence at craiyon.com. You type in some words and the AI creates nine pictures from the prompt.

Yesterday on my first go I was just trying to see what the AI could do. What were the edges of it’s ability? Could it create anything really interesting? Answer: yes.

Today I’m going to try to get it to draw interesting and expressive faces. How dramatic can we make them? Here’s my first try:

The face of a man overwhelmed by society

Wow, not bad. I’ve heard that Craiyon has difficulty with faces and they certainly are warped, but I like that. It makes them more dramatic and artistic. If these images were more realistic, like photographs, I don’t think they would be as interesting.

Let’s try some variations:

The face of a man overwhelmed by information

OK, not as good, but not bad. The “information” is kinda flat and covering too much of the face to see the emotion.

Next…

The face of a digital man overwhelmed

I was trying to get some sort of synthetic man under stress but ended up with heads with data projected on them.

Let’s try some women.

The face of a digital woman overwhelmed

Hmmm… That’s a little more interesting. Not sure what the red line at the top left is but there’s more emotion in these faces. I think changing the gender made a difference.

Who else can be overwhelmed?

The face of a cyborg overwhelmed

I’m liking this! What else can we try?

The face of a robot overwhelmed

Eh… It’s that cliched white plastic robot face. We lost almost all the emotion and it’s all the same bland generic robot. Except it’s not just a robot. It’s an ANDROID.

The face of an android overwhelmed

Oh no! Our corporate overlords are invading our art! Words change meaning over time. Context matters. Android isn’t a robot designed to emulate a human anymore. It’s a cartoon corporate mascot for a phone operating system! We’re way off track…

Tell the AI what to do!

The face of a cyborg crying

I love this thing! I could sit here all day typing variations!

The face of a cyborg overwhelmed with grief

I’m surprised that this “grief” render doesn’t have any crying. …well, maybe the bottom right.

The face of a cyborg overwhelmed with anger

No subtlety there…

The face of a cyborg overwhelmed with love

The heart – symbol of love.

The face of a cyborg overwhelmed with love in a painterly style

Now we’re talking! Let’s try this “painterly style” with other things.

The face of a male cyborg overwhelmed with love in a painterly style

Amazing…

The face of a man overwhelmed with grief in a painterly style

😯 There are no words…

The face of a woman overwhelmed with grief in a painterly style

The answer is… Yes. Artificial intelligence can draw emotion in a face, you just have to find the right words to prompt it. It’s part the ability of the AI and part your direction. Iteration is the key. If you stumble across a corporate mascot, back up and try another path.

Next time I’m going to try to create interesting faces giving the craiyon AI as little information as possible.

Artificial Intelligence – Can It Make Art?

Leave a reply

I’m checking out art created by artificial intelligence at craiyon.com. You type in some words and the AI creates nine pictures from the prompt. My goal here is not to create great art but to see what the AI can do. What are the edges of it’s ability? Can it create anything really interesting?

Here’s the first thing I tried:

Two people in love

Interesting, but I was expecting more. There’s two figures in most of the images but the abstract concept of “love” is mostly expressed by red hearts. The AI is not showing “love” in the art it’s showing a symbol of love.

Maybe I’ve already found something it can’t do? Does it take abstract concepts like “love” and convert them to simple symbolic images? These look like bland lazy power point clip-art. Lets try some more:

The human adventure is just beginning

OK, that’s better. I can see “adventure” expressed in most of the images. It’s mostly outdoor mountain climbing but that’s better than just simple heart symbols.

“The human adventure is just beginning” is the tagline to Star Trek – the Motion Picture, but I guess Craiyon didn’t get that at all. Searching on the same words in Google Images returns ONLY Star Trek pictures so this AI works very differently. It’s not just trolling the internet to find images that are related. There’s nothing in the Craion’s images to suggest Star Trek at all.

OK, let’s try something REALLY abstract…

It was the best of times, it was the worst of times

OK, I don’t know what to make of this at all. It’s interesting which is good but I’m not sure where any of this is coming from. Maybe “people in the news”?

Let’s try something more direct.

I make SciFi

OK… Looks good but all the pictures have exactly the same style of SciFi even down to the spaceship designs. It’s that generic SciFi style you see everywhere especially in games these days. Maybe I should type out “science fiction” instead of using an abbreviation.

I make science fiction

That’s pretty much exactly the same…

I make SciFi 50s

OK, there’s a different style, though I’d say it’s a little more 70s than 50s.

I make SciFi 70s

Hmmm… It’s not getting the style differences for the years. Maybe I should try robots?

Robot roll call

Now everything is a cartoon. I wonder how it determines the style if you aren’t specific about it. I’m going to just keep trying different stuff.

Brave New World

Wow! That’s interesting. There’s certainly some variety and emotion here. I guess you just have to hit on the right phrase for it to make a leap beyond the surface meaning of the words.

1984

Well it didn’t get the reference to the novel by Orwell (I didn’t think it would) but I guess it DID understand the number was supposed to be a year. All the images seem to be from a mid-eighties time period, I think.

I wonder what would happen if I input some song lyrics?

Stepping outside she is free

Well… Feet walking outside. These images didn’t pick up the poetry of the line at all.

Words are meaningless and forgettable

More symbols. “Words” = scrabble pieces. More clip-art for corporate power point presentations. The letters are unreadable but I don’t think that has anything to do with the phrase I typed in. I have heard that Craiyon has trouble with text.

a candy colored clown they call the sandman

Woah! Yep, those are Candy Colored Clowns. The direct straightforward descriptions yield predictable results.

The dreams in which I’m dying are the best I’ve ever had

Wow! That went somewhere interesting. I wasn’t expecting the drama in these images.

I wish I was a spaceman, the fastest guy alive

Oh well… Spaceman, that’s about it. I was at least expecting some of the images to reflect the speed aspect.

Expecting it to find the poetry of a phrase seems to be hit or miss. Maybe I should stop trying to throw it curve balls and be more descriptive and tell it specifically what I want.

SciFi adventure man Picaso style

Hmmm… OK. Sure… It’s a SciFi adventure man if he’s in a post-apocalyptic world. Not sure if this really looks like Picasso though. Oh wait I mis-spelled Picasso….

SciFi adventure man Picasso style

Woah! There’s Picasso! One letter off made a big difference! I think the SciFi got buried in the style though.

SciFi adventure man dali style

We’re making some art now! SciFi still means roaming the post apocalyptic desert apparently. Let’s try spelling out “science fiction” to see if it changes things.

science fiction adventure man dali style

Eh… Not as interesting. And apparently t-shirts are part of it now presumably because you can buy Dali’s art on t-shirts somewhere on the internet.

I’m going to try adding the word “face” to get a close-up of the man.

science fiction adventure man face dali style

OK, Now we’re getting just Dali’s face with hints of t-shirts. Craiyon is getting off track. I’m going to try limiting everything to faces next to see what happens. I’ll leave that to the next post though because this is already getting way too long.

Craiyon seems much more primitive than other AI art projects I have seen (but not tried) like Dall-E 2 and Midjourney. Those others seem to be able to create images that are much more realistic like they were made by professional artists. I bet they’re much more controllable too. That’s great but maybe not quite as interesting to me as Craiyon’s screwed up faces and unexpected results. Realistic is good, but I think interesting is better.

I’m concerned that as AI drawing technology gets better and better, all AI art might become more bland and predictable, just like what’s happened to most of the art on sites like ArtStation. A group-think of what is good and professional develops in the community that is then picked up and exaggerated by an algorithm that brings all the “best” to the top. Pretty soon everything looks exactly the same. Everyone is seeing and creating the same thing.

This is a special moment in the nascent development of AI art creation and we should take the time to appreciate the crazy and unexpected images for what they are, ART.

Check it out Craiyon for yourself.