BonFX | Typography & Graphic Design Blog

The, ahem, “League of AI Artists”, 2022 edition

October 12, 2022 by Douglas Bonneville

ai art dalle of an apple with an eye in it

“Apple of My Eye”, Douglas Bonneville using DALLE-2

(EDIT: Artist list already added to!)

AI art arrives, and is the apple of everyone’s eye. Well, not quite everyone.

There is no question that 2022 has been the year of the arrival of AI art (#aiart), or image synthesis (#imagesynthesis), or AI infused art (#aiia), or any of several other hashtags that have come to denote this new art form.

Mini-mini history lesson

There has been excitement, confusion, exploration, astonishment, advancement, rancor, scandal, and lots of dopamine. We barreled into 2022 riding the crest of Disco Diffusion mostly through Google Colab notebooks, which itself piggybacked on VQGAN through the fall and winter of 2021. And then all craziness broke loose in March 2022 with the arrival of DALL-E 2 which blew the doors off of everyone’s expectations of what diffusion based models would be capable of so quickly. And almost simultaneously, we saw an explosion in popularity of the Discord-based MidJourney and it’s novel take on diffusion, which took “beauty in AI art” to new levels. But DALL-E 2 was stealing all the air out of the room until August with the arrival of Stable Diffusion. Stable took everything we new about the possibilities of AI art, put it in a little box, and the blew the roof off the house with its open model.

A month later in September 2002, Stable diffusion had been integrated into Photoshop via 3rd party plugins, Canva, multiple Colab notebooks, and DreamStudio from Stability.AI, the maker of Stable Diffusion. Stable Diffusion as of writing this, is going everywhere all at once.

A move towards Stability

Huge advancements and innovations in inpainting, originally pioneered by DALL-E 2, have poured into Stable Diffusion from every direction. Since SD was released as open source, as well as making SD available through API for integration into any project anyone wanted to come up with, the pace of invention is currently speeding up and mushrooming throughout all the digital tools in art workflows across the planet.

A temporary controversy

“The Cat, The Horse, The Train” Douglas Bonneville with Stable Diffusion

As what I hope becomes a simple footnote in the history of AI art (a simple, short, boring one), there was a bit of controversy as to whether or not AI art is even art, whether or not it’s able to be copyrighted, and whether or not it will devour art jobs across the planet. The was controversy with some well-established artists, and controversy at the Colorado State Fair, and every major tech magazine has now published something on the topic. But hopefully everyone will have outgrown their Luddite-like proclivities by the end of the year (it’s already happening). That seems all but inevitable. The cat is out of the bag, the horse has fled the barn, and the train has left the station. Even the feathers from the pillow have been scattered into the winds. All at the same time. There is no going back.

This was the big year

It looks like 2022 will be (already is) the biggest watershed period in history since the invention of Photography. And nobody knows what is next, though some have some inklings about the possibilities.

Let’s name names

“Diverse Artists at a Party #1”, Douglas Bonneville with Stable Diffusion

But, what about the artists?

This article was not supposed to be a mini history lesson, but it turned out that way as much as I wanted to just list some artists and be done with it. We need just a little bit of context to understand exactly what is happening.

So what I’d like to do now is simply list artists who have established and maintained an active presence on Twitter, which seems to be the main place the earlier movers congregated, and where they post their art. A few AI art subreddits have also been very popular, but in a different way than Twitter. The twitter experience has been one of visuals, thought leadership, debate, argument, passion, beauty, ugliness, cliche, and plenty of ad hominem. The reddit communities have been more focused on the images, and less about the artists themselves, and brand building for those artists.

But that said, let’s get to brass tacks.

On July 28th, 2022 the Twitter user Surrealabs posted a serious (with a pinch of humor) take on the current lay of the AI artist landscape. Basically, they took all of the Twitter AI artists they were following and mapped them in a quadrant grid according to how they handled their prompts. The vectors were between creation methods: prompt-driven or process-driven, and then prompt privacy: secrecy or transparency. Here is that quadrant chart:

AI Artists 2022

But without the twitter handles, the original chart is a little less interesting. Enter Bokar N’Diaye and a lot of leg work to attach Twitter accounts to all the handle images. So as of a few months ago, here are all those labels in the next image:

ai artists

Personally, I like to think of this “snapshot” of names and places collected by one artists among many, to be kind of a candid Polaroid shot at a big party. Now it’s permanently attached by a large magnet to the refrigerator in the big kitchen of the Internet, forever!

Surrealabs has gone on to say that they are surprised that something created as a bit of fun and bit of a meme has grown such long legs!

I love that a meme I made that basically just showcases a mishmash of people I follow(ed) is somehow still a fixture in the community

— surea.i (@sureailabs) October 12, 2022

The big list of links

“Diverse Artists at a Party #3”, Douglas Bonneville with Stable Diffusion

And for fun, I’ve added direct links to the Twitter accounts. Not all these accounts may still be active. Some might be deactivated or taking a break, so your mileage may vary. But, here they are, for better or for worse, for posterity. Starting from the top left as best I can:

And here is a space just for the other artists who were in nearby latent space, just not in frame when the polaroid was taken:

https://twitter.com/deKxi

That’s it. That’s the list.

CODA…

“Diverse Artists at a Party #2”, Douglas Bonneville with Stable Diffusion

If you are reading this, I mean, if you read this far, you are seriously dedicated. I mean, I’m impressed. So kudos to you.

But, I want to say here, that if you know of any mistakes, omissions, exceptions, or anything else interesting to add of fact check, pu-lease let me know. You can get me at @dbonneville on Twitter, or add a comment to the tweet that brought you here so that others can see it too.

Thanks for reading!

Oh, BTW: AI Art is…

Art.

Even though…

I spent <10 min on the new art for this post.

How To Use DreamStudio / StableDiffusion To Create A Traditional Illustration

August 31, 2022 by Douglas Bonneville

(A micro micro micro-documentary with an all-caps title to make it look impressive) Let’s start with the finished illustration (you’ll want to zoom in on it, so click and read the small print):
1/10

After 1-2 hours of generating terrible concepts, I finally landed on the idea of a cat reading books on how do cat stuff, including manipulating humans. Fun! Started with maybe 30 “cats in chairs” renders in #stablediffusion to get ideas:
2/10

I liked the cat face in the first one but would need to hand edit the body to add a book so we could see the title. Then I decided I wanted him on a couch, not a chair, to make seeing his pile of books easier. But first I edited the prompt to give me library ideas:
3/10

Now I needed a couch. Did a quick google(!!) this time to even get my head around what I was trying to describe. First time in all my #promptography that I did that! Weird experience. Found and image, which led to a new prompt for a couch:
4/10

Composite time in #PhotoShop. Just needed to get all the elements on the screen in one shot, however bad. Rember, the goal is to create a reference to draw over. Here is what I came up with as the overall baseline concept. Rough but that is the point:
5/10

I’ve got a rough drawing now in good shape. The #cat got flipped horizontally and was mostly hidden by the book. Legs and tail added:
6/10

Let’s add some wash behind everything pen & ink to get that authentic look going:
7/10

Now, let’s push the background ink, but not the wash, further back to emphasize the foreground. I drew the library on a different layer than the couch so it was just a transparency setting in #procreate:
8/10

Final touch – let’s vignette it to give it that authentic New Yorker Cartoon vibe (the goal from the start) and we are done! #newyorkercartoons:
9/10

And that is the whole story! There are tools coming shortly (looking at you @artbreeder that could further optimize this process and make it more creative/exploratory at the same time. It’s a new dawn for #illustration creators…

DALL-E, pixel art, and a silly story

August 9, 2022 by Douglas Bonneville

Pixel art is very time consuming to create by hand, but with the advent of text-to-image AI tools like DALL-E, we can kind of cheat and get great output with very little effort.

The following images are from a silly story I quickly put together soon after getting DALL-E access in early 2022. On close inspection of any of these images, you’ll see where pixels sizes are implied rather than actual, and they kind of drift in places, get rounded corners in others. Just don’t look to close but still marvel at what DALL-E is able to create in seconds. Astonishing!

All concerns aside, this was just a very fun piece to work on, putting some prose together in an entertaining & comical story simply meant to showcase the ease with which DALL-E can create serviceable art.

So let’s get on with our story!

The New Kodank “Livestream Pix Cam”

(1/8) “CHECK THIS OUT! Kodank released their new Livestream pixel art camera “PIX CAM” with built-in 5G! Demo live in 3…2…1… OK it’s open, and on. How does this thing work? Ooops, we are streaming already!!…didn’t mean to take that!”

(2/8) “Let me try and read what the little thing here…oh shoot took another one. how am I doing this?”

(3/8) “Oh, I see. Capacitive picture button, ok. Let’s see…here is my lunch! Instagram eat your heart out…”

(4/8) “Now let’s flip this…oops DAAANNGG it just took a hard fall..C’mere you little thing…kinda slick there aren’t you…”

(5/8) “Let me try and show you what I’m working on, got a great new arti…HONEY! PERFECT TIMING geee I told you I was doing a shoot! Get going…go!”

(6/8) “Ok, we are going to get some good shots. Ah, here comes Felix! So cute right in my face, this will be a good shot…hold it Felix. Felix. Felix wait…oh Felix…”

(7/8) “Great, now there is someone at the door. I think it’s the UPS guy. Might have to sign for this BRB 5.”

(8/8) “I’m back, yeah had to sign for something expensive. K, last shot. Let me show you an apple tree in full bloom ready to harvest! 1…2…WWHHATTT!! What are the chances of getting a shot of lightning striking my tree!!! Gotta go!! Will post again ASAP!”

(9/8) “BONUS: Here is the box. My apple tree is gone forever but thanks for stopping by.”

FIN