intro_to_ai_art

The Definitive Guide to AI Art

I started this gallery because I wanted to build a personal collection of AI art in order to understand what exactly AI art was. I wanted to know how it was made, what was being made, and if it was truly art. If you have similar questions this is a good starting place for you. If you already know the answers - awesome! - this should still be a fun overview as I'll walk through the best of what I bought. Experienced AI artists and collectors will see some familiar pieces/artists, but also a few you've probably never seen before.

Note: there is so much good art of all types being created in the field of AI - but in this gallery I'm going to focus exclusively on AI art being created as digital tokens (or NFTs) that can be created, bought, and sold on sites such as objkt, opensea, etc.

As you can see from the examples so far, it's pretty amazing what AI can create. You've probably seen examples on social media of what can be created from a simple prompt using programs like DALL-E Mini. For example, typing a phrase such as "a painting of a snowy night in the style of edward hopper" can create a surprisingly astute image such as this:

Click here to hear this Holly+ remix

So what is AI art? I define it as art where AI - in any way - is involved in the creation of the work. Similar to generative art, AI art is a partnership between computer (in this case, the AI) and human. In the best of AI art these two work together in harmony.

AI art can look like a painting or a picture (like the above). It can be animated. It can be a video (like the piece at the very top). It can look like you are entering the metaverse or it can be black and white, grainy and glitchy. It can look digital or analog. It can even be music or text. It's all AI art.

Just this "prompting" (entering text) of the AI - and nothing else - can be the entire process of an artist. Some artists focus on iterating on these prompts, making prompting an art in itself, learning from the AI and instructing the AI until they create something beautiful.

Of course, many artists use the prompt as an initial step. Some combine the best of several outputs. Some take the output and post-process it - adding their own touch, adding texture, altering palettes, etc. Some artists create custom AI programs, "teaching" it with a series of carefully curated images that will produce a custom look. Some artists feed their original digital paintings into the AI as a base to start from.

In the painting above, xeronimo used the raw ai created by the prompt with no post-processing.

Of course, one of the first questions many people ask, and that I mentioned above, is that if anyone, with just a few words and a few minutes, can create something so complex and artistic - is that still art?

If there is no "skill" left in art - if you don't need to know how to actually paint (or draw, or whatever) - if you don't need to practice for years learning brush techniques and color theory - if anyone can create these images - then what exactly is art now? Are we redefining art?

Also - hang on - because in the middle of all this I'm going to create some AI art myself. To see how easy it is, how difficult it is, and if I can actually create something worth looking at.

Inevitably AI, in conjunction with physical hardware, will be able to actually create sculptures, actually paint the canvas, etc. And of course it can already write poems and stories, news reports, make videos, make movies. So while we are currently limited to images on a screen - this question of "what is art" is going to soon expand to all art forms.

Perhaps art becomes taste and curation, like the above flowers, where the AI creates the art, but the artist has tastes that are similar to your taste, and experiments with prompts until he/she produces something - in this case a collection of flowers - that you think is beautiful.

Perhaps art becomes what the artists does in post-processing - combining images, adding effects, digitally painting, etc.

Perhaps art becomes vision - an artist has a unique and glorious vision and works until the AI creates it. Maybe the art is the skill in creating prompts that result in that vision.

by @ragnar_meta

(Best viewed on a big screen. Don't use a phone)

ATNPassion, ELEMENTS #453

Let's try to answer some of them while we look at the art itself. We'll take a tour of AI art using my collection as examples. We'll see a variety of more traditional painting styles, but we'll also see examples of other styles - videos, illustrations, collages, music, and more. We'll see how to create AI art and some of the software involved. We'll even get a little philosophical at the end.

Let's start with something beautiful and representative of much of the AI art that has been created in the first half of 2022.

Ok - before we go too far into the collection, I want to pause for a quick moment so that I can confess. And this is important and relates to the big questions around "what is art?"

Collecting AI art was difficult for me at first. I was very cautious - even nervous - about creating this gallery. Because, as we said, with AI anyone can create beautiful AI art. You just type a sentence into a website and boom - you have incredible images that in prior years would have taken an artist decades, even a lifetime, of practice to create.

And now, a five-year-old can do it. (my five-year-old did!)

So what happens when I pay a bunch of money for (and showcase) a piece that some anonymous kid in his bedroom created in about 2 minutes by typing a few well-chosen words into a website ... and everyone is laughing at me for picking it and are now closing this tab...

But what happens when I see a piece by a brand-new artist that no one has seen. And I don't know if the artist spent weeks refining and working on the piece - which many of the best artists do - or if the artist took 2 minutes on his lunch break.

I know better than this, of course. I know to buy art that I love, regardless. And that, as with much modern art, sometimes the value of the piece is really in the meaning or the backstory. But that's easy to say, much harder to do when you're spending money and putting your choices on display. The ease of creating AI art made me cautious - plus, I don't want to spend money on someone who barely even tried, right?

What do you think?

That was a lot of questions. We'll explore them more later on. But for now, to form our answers, I think we need to start really looking at the art.

Let's start with three examples of world building, where the artist focuses on creating a unique world over a collection of works.

Let's move from world building to styles.

Here are several pieces by RenAI_NFT who creates art in the style of classical renaissance art.

His process is typical of many AI artists. (For details click here)

Portraits are a major genre within AI art - I think it's fascinating to see how AI interprets humans. Here are some AI+human views of the world through portraiture.

I like BrainDrops because they curate an excellent representation across the spectrum of AI art, varied and always interesting. I'll show one of each of their collections here. Head over to braindrops to learn more.

Because so much of AI art leans dark, finding something whimsical can be a relief, such as with these pieces by Elena Lazutina.

Surprisingly, I found myself attracted to art involving flowers. And AI art has a lot of flowers!

Xeronimo is a great example of "taste as art." He captures loneliness like no one else I've seen in AI art. If you dig lonely, sad paintings, you're gonna' like what he creates. I get the same deep loneliness from his works that I get from Hopper.

Here is his series, "SuperMarket Loneliness."

OK back to the art. Let's get to it now and look through a ton of art you might find interesting.

A lot of AI art leans towards the grotesque. Is this how AI sees humans? Or how we see ourselves? Apparently it is.

Here are three pieces by NeuralDayDream. Maybe a little Francis Bacon influence here; pain, suffering, body parts.

But that grotesque, when mastered, can represent something non-grotesque - such as erotic, as in these pieces by Ilya Shkipin.

What I like about Ilya is his artistic ability to use AI to create emotional responses. The evocative details on the piece on the left (aside from the, ahem, suggestive pose) such as knee-high socks, the hand behind the head, the smirk (but even where is that smirk?). And the image on the right - so abstract we don't even know what we're looking at when we look at the details, but as a whole when we step back ... it may be short on details and blurry and abstract, but it's nearly pornographic, isn't it?

On that note above about a five-year-old being able to create AI art ... I think it's time I gave it a try. It'll not only demonstrate how AI art is made (or at least one way it can be made) and what AI can do - but also give us a foundation to appreciate (or not appreciate) what others have created.

If you already know how all this is done, feel free to skip ahead to more art!

OK - that adventure is over for now! Back to the collection now that we have even more context.

And that's a good segue into music+AI NFTs.

There aren't very many of these, though there is a lot of work being done in this area. Just not much available yet as NFTs.

The first one from the holly+ project by Holly Herndon. Herdon recorded her voice - then trained an AI voice model with it - and released it to artists to use in their own songs. The AI can replicate Holly's voice for custom songs. This one is a Ronald F. Bustamante remix of a Mahler song already AI remixed by Satiearranged. (the image isn't rendering on this one below - click through).

The second one, Waveforms #7 by neuralnet, is a set of waveforms generated using a neural network.

Somewhat related to history, let me give a brief note about decentralization in NFT art, or the lack of any central authority that sells, stores, or controls in any way the artwork.

One of the great benefits of NFT art - and AI art on NFTs - is that artists have a full freedom to create. There is very little (or no) control over the market- anyone can create anything. (If you want more info on what decentralization is, there's a lot of information out there. We won't cover it here.)

It doesn't matter who you are in a decentralized art world - no one can stop you. You can be a pretentious artist, a down to earth artist, not an artist, old, young, poor, rich, a nobody, or someone famous - in the NFT/decentralized world, if the art is good, it's good.

Art used to be a vocation. A career where you could make a living. But in the last century, because of the lack of livable wages in art, art has become a hobby for most people. It's now reserved for the few who are good at the art  the networking. And even then, very, very few artists make enough money to live. Art simply is not a career option for most smart, resourceful people, even if they are talented and interested. We no longer have masses of smart, creative people dedicated to exploring art. We've lost a century of creativity this way.

I could be wrong (and if I am, let me know), but the oldest I could find was AI Generated Nude Portrait #1 minted on Ethereum on April 5, 2018 by Robbie Barrat who, by the way, is aware of its uniqueness.

The oldest I could find on Tezos is Amorsolo Dream No. 1 minted on March 1, 2021 by the now disappeared sintang_ligalig.

The above work, "Art is art" by Nikita, was released the day after DALL-E 2 (the current leading AI program) opened up the ability for artists to sell commercial works based on its AI (July 20, 2022). Many artists are currently exploring this question of "what is art" with their own AI art. I think Nikita is one of the masters of this new AI art and this piece will become emblematic of this period of disturbance in art.

There are many questions, and many strong opinions, about AI art. For example, if you just "tell" the AI what to create, and it creates it - is that still art? Or does a human need to alter it somehow to be true art? And if anyone can easily make beautiful art with AI - are we now all accomplished artists? Do we need technical skills to be artists? Do artists have to prove effort in order to be valued?

These are valid, and difficult, questions.

Now let's step into portraits.

Next let's take a brief journey into the history of AI nfts.

Let's move back into more painting style with a few more from my collection that I think are worth looking at. I won't offer much commentary. Just scroll through, take your time, and enjoy!

After spending hundreds of hours looking through thousands of pieces of AI art to assemble this essay/gallery, I’d like to say a couple things about AI and art, and what I think the answers are to a few of those questions we asked. I’ll be brief. But I think it’s important.

Sometimes the result is surprising - even the same prompt given twice produces different results. Often what the AI returns is nowhere close to what you expected. So you refine the prompt and try again. And again. And again. It's a back and forth of creativity and interpretation. Until eventually, you get something close to what you were looking for. This process can take hours, or even days.

You can see this is all about iterations. Refining, interpreting, adjusting, being surprised, trying variations ... until you see something you like. Some call this "prompt engineering". There are even bots that can help you create prompts.

For my project, I decided to create "alternate" book covers for a few novels that made a difference in my life (some when I was younger, some more recently). I had a few goals: I wanted to create a variety of covers while establishing a consistent style; I wanted to create images that didn't look like the current/famous covers yet still captured the spirit of the book; and I wanted to show off what AI+human can do.

Remember I was a total beginner going into this project. And I'm not a visual artist.

For my project, I tried out several AI programs - NightCafe studio, Craiion, artbreeder, prosepainter - but settled on MidJourney, which seems to be the tool of choice for many artists. DALL-E 2 is the top choice, but it takes a few months to get off the waiting list and get access, and I didn't have that much time.

The process of concept to NFT ready to sell consists of 3 steps. 1 - create the AI image 2 - post-production 3 - create the NFT

Next I picked out my novels and got to work.

I'd say it took about 10-15 hours of work to get the 14 outputs I ended up with. It became easier as I learned about prompting. After a ton of iterating and saving images, I was ready to move to the next step.

Here's my favorite cover - one for The Things They Carried by Tim O'Brien, a novel about Vietnam made up of a dozen or so shorter pieces, tied together, and based on true stories, but not true.

I ended up with a few versions I liked (shown on the right), but this final one I kept with no post-production at all. I love the way it captured my favorite scene in the book - one that is repeated many times - of a friend of the author stepping on a land mine and floating up into the air and into a tree.

Once you have the output(s) you want - you're either done (what's called "raw" AI), or you head into post-processing, where you add your own touches. Some artists are heavy on the post-production, some are light, some do none.

In my case, it was pretty light. I just cleaned up a few elements, using photoshop, that I thought were distracting. But I kept 98% of the raw AI output.

For The Remains of the Day, I did a little post-processing, as the structure behind the man looked like a castle when I wanted an English Manor, and his clothes looked a little too "monk-like." So I made some minor adjustments. I lopped off a few turrets and smoothed out the cloak.

I could go on for quite some time about why I love each of these covers, and what each captures about the corresponding novel, but that's not really the point here.

Here's the collection as a whole after step 2:

The next and final step is to mint the art as NFTs. I did this through objkt.com - it was very easy. I won't walk through the steps here (there are many tutorials) - but it only took a couple minutes for each one. You can then decide how many copies of each will be available (many pieces are in collections of 10 or even 100) and set the price.

And so my collection of alternative book covers is live! The first NFTs that I created. I recommend you create some yourself.

If at the end of this gallery, you found this information useful, feel free to click here and hop over to objkt and buy one. :) They'll help support the creation of the next gallery (I don't get paid for these).

(If you're up for a challenge, and know a little about books, see if you can guess what each novel is (you can see the full list here - a few are a little obscure so don't get frustrated) and what style I told the AI to use (answer at the end of the gallery).

Pre post-production with turrets and monk clothes

Two alternate final covers

Step 1: Create the AI image

To create the initial image, you give the AI what's called a "prompt" - or, a description of what you want it to generate. Then you let the AI run and interpret your words.

Step 2: Post-Production

Step 3: Create NFTs

For my prompts for the real project, I tried not to use the title of the book - that's too easy and most of the time when I did use the title the output just looked like the current book cover (which makes sense). Instead, I tried to explain my favorite scene, add in details and emotions I remember from the book, that kind of thing. And I wanted all the covers to match with a holistic style.

The learning curve was actually pretty short. I spent about a full day playing with the AI learning how to use prompts, watching what prompts others used, looking at outputs, etc, before I was ready to start.

For example, here is one of the results returned for "off center still life of a flower bouquet in a glass jar in front a window with sunlight streaming through painted in a loose painterly style"

This piece above, Vanitas V by xeronimo, is a perfect representative piece.

First, importantly, I think it's an accomplished work. xeronimo is an artist with an eye for detail.

But beyond that, this piece is representative for several reasons.

First, at first glance, the piece like much of AI art is ... a little weird. There are almost concrete figures in the piece - but when you look closely at the details, they disappear. Something is off. And it's off in the same way Picasso's works are off, in how he takes a form and abstracts it, breaks it down into components. What we see here is similar to that, except Picasso often took the pieces from the same subject - a woman's body for example. Here, the pieces are taken from disparate places and then placed together in some sort of strange collage that somehow works because the AI knows it is all related. And you get the feeling of what the artist wants you to feel.

Also this piece has several other representative qualities of much of the AI art being created:

A quick word about the "blurry" weirdness of faces in most of AI art today. As mentioned earlier, this is mostly because of limitations in AI. It isn't perfect yet. But it will be (see example on the left of what the newer AIs can create).

I like jonCates's thoughts on this in his series of tweets

"AI Art ambiguity is specific to our era rn, in this development phase of nonhuman systems ... [I] feel that these current AI Glitch Art opportunities are prolly going to expire and become more difficult to capture/create in near futures as resolutions scale

"oc folks will try/build systems to create the look/feel of time-specific AI/ML models ie a 'June 2022 MJ' @midjourney aesthetic etc, however, their attempts will be bound by computing, by codesources that are {open|closed}, [blackboxed/proprietary], +/or [shared/sold]

One aspect you may notice about AI art is a dali-esque blurriness of faces and other details. We'll hear a few words on this later on from Jon Cates (who created the western series above) when we get to portraits. But for now, you might view this common aspect as a combination of limitations of AI + an affinity of current AI artists for glitch or blur style art.

The last example from BrainDrops involves music (don't forget to turn on your sound). Fake Feelings by Dadabots x Silverstein use a 2-channel stereo neural synthesis model to generate audio in the style of post-hardcore emo, trained on the music of Silverstein. There are no lyrics. Just random syllables!

(above, we see the first collection launched by BrainDrops, the podGANs by Van Arman. These use AI in their algorithms to create high-res GANs using low-res 8bit art.)

Elements by [ATNPassion](https://www.atnpassion.com] are a collection of AI-generated fantasy worlds inspired by interpretations of earth, air, water, and fire.

Genesis is a collection of raw AI images by Claire Silver that have been created and arranged to tell a story.

Silver is one of the leading figures of newer AI artwork, often speaks about AI art, and is really, really good at what she does. She is one of the masters.

Chimerical Stories by Sofia Crespo and Feileacan McCormick is a series of images that crossbreed jellyfish-esque aspects with butterflies to create new species of chimera.

First, a series by jonCates, "mechanicalInterventions: The Artificially Illustrated Glitch Western Primer for Machine Learnerrs." I love his eye for combining glitch style with AI with westerns.

Wayfinder has built a world of a traveler, The Wayfinder, discovering the people and places in alternate realities.

Din Burns has built a dystopian universe that is "something between our present stats of the world and out possible dark future." His work is dark, colorful, and dream-like.

One interesting freedom that AI gives to artists is the ability to explore multiple genres. Since you can pick most any style when generating the initial AI output - today I want to be a cubist! - artists have the freedom to explore all styles, as we see below from xeronimo.

Nikita, seen in many pieces in this gallery, is in my opinion, among the absolute the best AI artists. Here is his "fragile" series where he creates portraits of young people "where the fragility is visible, yet they are self-sufficient and beautiful."

Let's start with two portraits from "When we were kings," by Alonerone. Alonerone explains that his work is "a dialogue between urban reality and forgotten orientalism"

The earliest one I can afford is the Mario Klingemann just above from April, 2021. Mario is one of the "old breed' AI artists who have been experimenting and creating art with AI for years.

And this collection of AI dolls from Kaykayok.

Maneki Neko creates incredibly intricate collages of flowers (and she creates owls!). Notice on the left how sometimes there is just the impression of a real flower, but upon inspection the flower isn't really there.

And there's a person hidden inside!

It's a little odd to talk about the history of AI since it's so new. Mostly we are, as they say, making history right now. There aren't yet any so called "grails." (I'd make the case that Nikita's "Art is art" is a grail - but we'll just leave that here.

There are, however, a few interesting ideas to explore. First, what are the earliest NFTs minted involving AI art?

Now as far as I know, Hopper never painted a snowy scene. But the AI figured it out.

Maybe art becomes an act of storytelling such as the above series - how an artist puts together a series of images to create worlds, tell us a story about that world, and in the process moves us with the pacing, the flow, the vision, the visuals.

Maybe the artist combines all of these, and through months of work, creates a prompt, that in conjunction with how an AI understands humans and post-processing skills of the artists, creates something we've never seen. Something new and fresh. Something that is art.

Mario Klingemann, Headwinds (1/500) Klingemann is a pioneer - an OG - of AI art

Eventually, after looking at a couple thousand pieces of AI art, I started to see patterns, I started to get a feel for what was easy, what was unique. And I started to spot - at least what I thought - was great art. And of course, the ultimate truth still is: do I like the art. Do I connect with the artist's view of the world and how they choose to present it, and does that view resonate with me in a unique and non-cliched way. Does the work create in me an emotion that fires off that dopamine. If it does, I should like it without pause. But still...

Thomas Kincaid is art, right?

Alonerone, Wounds II (1/30)

xeronimo (Jerome Herr), Vanitas V (1/1)

Nikita, Art is art (1/1)

KaykayOhhhK (laalis1776), Paper Plantae

#44

#110

#16

#8

introvoid, Spectre Ignition (1/120)

clownvamp, The Flowers of Hyde Park (1/26)

RenAIssance, An Evening in Belagio (1/3)

RenAIssance, A Dead Knight (1/12)

Clockwise from top left, all by Wayfinder, Ghost (1/12), Encounters (1/10), Outsider (1/15), and The Fates (1/8).

Clockwise from top left, all by Din Birns, Latent Space (1/10), The Tower (1/5), Seeking Forever (1/1), and Interconnected (1/15)

xeronimo, SuperMarket Loneliness II (1/10)

xeronimo, SuperMarket Loneliness I (1/10)

xeronimo, SuperMarket Loneliness III (1/10)

Clockwise from top: [ILLUSTRATION] #8q4314U, #4h4922B, #1r8090E, #3t7298M, #5i5667H, #7a2587C - all (1/10)

A series by RedruM and RedruM+FlexasaurusRex

DVKtheartist, A sunny morning after a rainy night (1/1)

Alonerone, "Q" (1/20)

Alonerone, "Autodafe II" (1/20)

Nikita, "fragile #78" (1/1)

Nikita, "fragile" (1/39)

Dehiscence (Ghosts in the Light), "Ghosts - 003" (1/10)

Henry Daubrez (Upskydown), "Terra Nova" (1/15)

Meloman, Persona (1/20)

Nude Robot, V3GA Outcast (1/25)

Ilya Shkipin, Ghost of Sleep Paralysis (1/20)

Thomas Intuitive Art, AI Faces #4

Thomas Intuitive Art, AI Faces #5

Dani Leoni, Tiles (1/1)

robness, ARTIFICIAL TRASHBAG SELF PORTRAIT OF ROBNESS_V2, 1/1000

Ganchitecture, Byblos - The Citidel (1/20)

Bruta, Open you eyes (1/20)

In the Elements series by WetPotatoBrain a model is trained based on the elements of art - color, space, shape, etc. It is then asked to create variations. These three are based on shape.

Memo Akten, autopoeisic transmorgification fragment #001 and #003. 30 seconds of ... something omg. With audio. This piece was created with custom software using AI. (sound on)

A newer piece by Nikita showing the realism that DALLE-2 is capable of.

0009, in·stal·la·tion (1/20)

Nikita, steel and skin #2 (1/10)

That's directly related to our question -what is art - and more specifically, does effort = value.

I felt this deeply and wanted to curate well. But AI art is very new - and there isn't a lot of criticism to draw from. If you want to judge it, you'll have to do a lot of original thinking. So at first, I only bought art that someone else had validated. Pieces intelligent people love or have been featured in a gallery, etc.

For example, this piece by Ilya Shkipin. Ilya is highly respected in the community and this particular piece was just featured in a gallery. Others have judged it for me.

Again: does effort=value?

MidJourney returns four variations from the AI, which you can then keep or ask it or create further variations from. Or, often, you don't like output, refine your prompt, and start over. It was very difficult to get that vision in my head translated to the AI through prompts.

Dani Leoni is another of my favorites. Her process is to let the AI create the initial image, then perfect the piece with heavy post-processing using her own style of digital painting.

These AI flowers in the Latent Garden series by PixelFiller are generated real-time in the browser by GANNN. They might take a few seconds to render.

And a few more plants...

I don't know where to put these next few, but they are pretty cool and different, and they sort of fit in here.

prettybad, every day begins to feel the same (1/15)

Ilya Shkipin, Sleepwalker 91 (1/1)

Ilya Shkipin, A lost ticket (1/1)

Introvoid, The Kiss (1/20)

nvnot_, birthplace of creation (1/10)

Jeremy J Torman, Interdimensional Time Travel (1/4)

Phosphor, feed addiction (1/25)

Isolationist, Aware of Being Aware (1/10)

Two more stunning pieces by Dani Leoni. I can't say enough about her work. She is at the top of her craft.

Dani Leoni, Ojos cerrados (1/1)

Dani Leoni, Swan II (1/1)

Dani Leoni, Festin (1/25)

This portrait is not simple. It's not an easy "mugshot" like you might expect from a machine. This portrait has motion and sensibility. It looks like a selfie, or a photo capturing an intense moment. It's not just realistic - it's natural in the way a human really would pose.

Aside from the message of these pieces, what I notice is how frighteningly well the AI understands the subtle clues humans need to form concepts. Yes, the artist may have aided in this in post-processing, but it's also being generated from the AI.

Above on the left, for example, notice the woman and daughter. We assume, subconsciously, that we see a woman and daughter. And we assume this because of visual clues - they have matching body language, mimicking each other, the same slouch, the same body shape. Their proximity that tells us they know each other well. An artist understands these details and uses them in the creation of the piece - but we as viewers tend to just absorb the details without thinking as part of our emotional reaction to the piece.

In the piece to the right, another by Nikita, look at the way the hand of the woman slightly touches the robot, in an eastern, formal movement; the pose of the robot kissing and how it's a natural "caught in motion snaphot"; how the emotions on the faces are reflected in the bodies. These are nuances that guide your reaction to the piece, and that most people aren't aware of, we just know when they are wrong. They aren't wrong here.

Yankel Vision, This Is Not A Work Of Glitch Art (1/25)

In a play on The Treachery of Images (Ceci n’est pas une pipe) by Magritte. "It is a representation of a glitch, is it not? The machine, you could say, is working just perfectly..."

I asked a few of the artists with works featured in this gallery.

Gene Kogan, Brain Loops - a collection of looping videos, each uniquely made from a combination of various subjects and artistic genres.

Xander Steenbrugge, Love at first sight from the DreamScapes collection - "an attempt to recreate the fantastical dreamworlds we all experience every night through the lens of AI"

Devi Parikh, Confluence - pieces are generated using a neural generative model (AI) trained on the artist's physical sketchbook.

Vadim Epstein x COH, TEOPEMA - 696 unique pieces; each one is a 3 sec video with sound. When all the pieces are put together in order, they create a 35 minute long exhibition video of a continuous deep dive into a world.

Artemis, SIGHTS - A generated visual tour of a machine uncovering a diverse universe of shapes and colors.

Nikolay Jetchev , ClipMatrix Creatures - 3D sculpting and texturing with words

Obvious, Dream Capsules - 1000 dreams created using Artificial Intelligence

An Introduction

1/ Art is Art

2/ World Building + Style Exploring

3/ A Confession

4/ I'll do it myself (an exploration)

5/ Curation and Where to Buy

6/ Portraits

7/ (a sidestep into a brief history)

8/ Whole lotta art

9/ A Conclusion (let's get philosophical)

So you love the art now and want to buy some! Where exactly do you buy it?

There are several well-known sites where the art is available - fxhash, objkt, opensea, versum, and a few others. These markets allow artists to upload their art, offer the art as NFTs, and allow buying/selling/offers/auctions on the art. The good news: compared to many other genres, AI art is still affordable. Most of these pieces cost me under $100, many just a few dollars. (There are expensive pieces out there, of course, that can be tens of thousands or more.)

One quick note: be sure to notice that many AI pieces are offered in editions (similar to traditional art prints). Often you are buying one copy of the art out of 10, or 100, or even 1000 identical copies. If the piece is the only copy, it is called a 1/1.

With so much AI art being created (you know, as we said, it's pretty easy to create) - curation is becoming very important. Most people don't have time to look through a thousand mediocre pieces to find that one diamond.

There are several established curation sites that have made a name for themselves as curators of AI art. Two of my favorites are Mirage Gallery and BrainDrops. I can't get the pieces from Mirage to render here, so unfortunately can't show those. But we can walk through the BrainDrops collection.

Of course, we are in the middle of a second transition now (in august 2022) with the commercial availability of works made with dall-e 2.

• Start with a concept

• Research what's already been done - colors, styles, composition, etc.

• Start prompting. Over, and over, and over. Tweaking the prompt continuously.

• A few hours/days of this, you'll have a few favorites saved

• Pick one as a foundation, add in pieces from the others

• Edit - point, zoom, snap, erase, etc

• Upsample to a large image (using AI again)

• Work on the fine details - curves, tones, colors

• The final piece!

Here are a few examples from when I was iterating to reach the above.

glass jar with flowers inside placed in front of a window with light streaming through the window in a peaceful mood in a painterly style with muted colors tuscan

a painting of flowers in a jar in front of a window

offcenter flowers in a glass jar in front a window with sunlight streaming in as a painting

it actually made it a "painting" this time

flowers inside the jars!

Here are a few other artists' comments on AI art, and the meeting of the new and old, AI and human.

So given all that - what does art become?

Here are a variety of portraits that I find interesting, unique, or just fun. See what you like!

jonCates, [∅.Ƒ31] (1/1)

Petra Voice, afterglow #10 (1/10)

It's kind of funny to say "old breed" since AI art is so new. Still, there was a transition in style somewhere in 2021.

Before then, artists such as Klingemann (above) and Memo Akten (below) typically used custom-built AI and were often from research or educational backgrounds.

Somewhere in 2021 with the release of commercial tools such as MidJourney and Dall-E we moved to a new style as we see in most of this gallery.

AI art is often serious and dark. I'm not sure why it leans this way - maybe because artists on the cutting edge of science are dystopian? This piece by Ganchitecture who focuses on AI architecture reminds me, probably obviously, of 9/11 and the twin towers, and especially of the cover of Basinski's Disintegration Loops.

These two from the Deep Sketch series by Bálint Kulcsár. These use an AI model trained on human hand drawings. The model uses a 'Type' (Cat, Bird, Tractor, Skull, Ambulance or Pig) and when the art loads, it attempts to draw that type! The art will be different every time you reload the NFT (they take a moment to render).

First - yes, AI art of course is art. But it hasn’t arrived yet. It’s still an infant genre mostly made up of highly-derivative works. Which is understandable and expected. We’re living through the first steps of a new genre and just starting to move away from traditions. Some of the current artists are good - really, really good. And some of the works of art are beautiful and nuanced and maybe someday will become grails of this movement. Maybe some of those future grails have been collected here.

Soon some artist - maybe one working today, maybe one just getting started - will take advantage of all that AI can do, create more than just what’s an obvious extension of the old world, and blow us all away. I don’t know what that looks like. If I did, I’d create it. That’s the point. Right now, as the artist @0DDxJ0BS says “We're not even on Page One."

But there's a new world of art coming from AI and there’s no stopping it. We’re going to see all the cliches of a new form - unexpected beauty, shock pieces, and as much AI/VR porn as the world could ever need. AI is about to change all art, not just painting. It’s coming to screenplays, books, movies. Whatever your art, it will be there. I can’t stress this enough. It’s already happening. It’s already here. You can try to fight it, lamenting about the way it used to be, saying it will never happen NEVER, how art needs to be tactile, hands-on, made by a human in order to really mean something. The whole old man shakes first at cloud meme.

Don’t worry - there are still markets today for the realists and the pop artists and the pointillists. And probably there always will be. But art moves on…

But we’re not there yet

My second conclusion? Craftsmanship is dead.

Think back to the above where I created my book covers. Truly anyone can do this. You don't have to be good at the mechanics of art - you don't have to spend a decade of your youth becoming good at crafting sentences. Or at brush work. Or at sculpting. You don't need it. You do now, you won't soon. Unless you just want to. That's fine. But unnecessary.

Claire Silver likes to say that now with AI, “Art is taste.” But I disagree. Art, I believe, is becoming thought - the ability to reflect, to formulate insights into our world that are worth communicating, the ability to project those thoughts in a concise and intelligible manner.

Yes, art already involves all those things. Art at one time was mostly utilitarian - the church commissioned pieces to get people into church and drive home moments from the bible. Artists did their best to make them beautiful, but they did what they were told. Then art became a way to express feelings and comment on society. The artist became a skilled craftsman who was also skilled at communication and commentary.

But with AI, we no longer need that first part - craftsmanship. We no longer need physical talents of any kind. The medium is now cognition. Art is becoming purely the ability of someone to share their world view through the AI. I don’t care if you make movies, write books, or paint with oil. Your art is moving exclusively to the cerebral - expressing your world-view, expressing your thoughts, your inner life, to the AI in a way that allows others to experience it. Where before that was the goal and the craftmanship was the method. Now cognitive is both.

Check Underneath (1/5)

Substitute (1/5)

Together Forever (1/1)

Hard Candy (1/31)

Static X (1/87)

25 Stripe St, (1/12)

28 Chocolate St. (1/12)

Doll Deluxe (1/5)

Doll with Owl (1/5)

Perplexing (1/1)

Irisated (1/23)

The Wisdom of Purple (1/42)

Flower Moonbird #17

Flower Moonbird #16

TheToshio, WONDERFUNGI (1/10)

TheToshio, ASENSIA (1/10)

Jenia Filatova, creature 👻🐨 (1/1)

Jenia Filatova, creature🍓(1/1)

Elements #443

Elements #483

Elements #125

Dani Leoni, Inner garden (1/1)

Nikita, special #2 (1/1)

neuralnet, Waveforms #7

drifting cuddlefish #333

amorphous nautiloid #333

Decoy Jealousy - Listen Here

Heavens Last Angel, Deep Journeys - each piece is an audiovisual voyage scored with original music designed to enhance the immersive psychedelic experience.

Nikita: "Even if a random kid with an AI and a little luck creates something iconic that touches millions of people - that would be great art ... but doesn't make that kid a great artist."

Jon Cates had two interesting points:

jonCates (I'm paraphrasing):

First - we have thousands of years of culture that have shaped our opinion of effort=value. It won't go away soon.

And second - there is great effort here - on the part of the programmers and crafters that created the AI that is helping to create this art.

xeronimo, FL_08

xeronimo, FL_05

Claire Silver, Page 77

Claire Silver, Page 100

Claire Silver, Page 275

Be sure to also check out my Definitive Guide to Gen Art