From AI Artist to cryptoArtist : My Five Year Journey

It took me five years to become a cryptoArtist. I have long called myself a cryptoArtist, but now realize I probably wasn’t until just recently. This is my journey from entering the space as an AI Artist that was using crypto to SELL art, to becoming a cryptoArtist that was using the blockchain to MAKE art. There is a big difference.

This month in 2018, I minted my first work on SuperRare. It was a photograph of one of my robot painted portraits. My next several mints were similar robot paintings, several of them part of what has become known as my AI Imagined Faces. (See image below left which was recently accepted into LACMAs permanent collection) They were popular, but I had a personal problem with them. They weren’t truly digital art. They were photographs of paintings. This always made me feel like an imposter in the space. Felt this way for years because often I was creating physical paintings which where not natively digital, but then I would mint them and present them as digital art.

The imposter syndrome has faded however with some of my recent releases including AI Imagined Faces : On-Chain (See image below right). This work is so digital, that it doesn’t even have a file. It is 100% contained on the blockchain, with no IPFS link, no arweave, no external references. The significance of 100% on-chain art is that the medium matters and if you use the blockchain as a medium, there should be a reason for it. At the core of my five year journey to becoming a cryptoArtist is the realization that unless there is a reason to mint on the blockchain, one shouldn’t. But to understand this journey, we have to go to the beginning.

Before cryptoArt:

Have been making art with AI for almost 20 years. My first robot was built sometime around 2005. I built it in my basement as an assistant to help me apply paint to a canvas with brushes. Over the next decade I programmed and built multiple improvements into multiple new robots. Over this time, I had a TEDx Talk, appeared in the documentary MACHINE - The Art of Intelligence, and got covered by a whole bunch of publications. Possibly the highpoint of it all was being awarded First Place in the international Robot Art 2018 competition.

Much of the attention I was getting was for my AI Imagined Faces. The AI was simple and used just two cooperating algorithms. The first would run a GAN to imagine a face and paint it stroke by stroke on a canvas. As it was painting, a second facial recognition algorithm was taking photos of the canvas and attempting to detect a face. As soon as it saw one, it stopped the robot from painting. The result was a haunting portrait that sat right on the edge of being recognizable as a human face.

We painted 128 AI Imagined faces on 9”x12” canvas board, 128 on 16”x20” paper, and about a dozen on stretched canvas. The SuperRare 1/1s were on stretched canvas. Interestingly, in addition to minting on SuperRare in 2018, I also recorded the 128 16”20” paintings on BTC BlockChain with Verisart. These were early uses of the blockchain to record the provenance of the art, but in my mind it was little more than that. All I was using the blockchain for was helping with traditional art sales of a traditional medium. To me, this did not feel like cryptoArt, even though I was calling myself a cryptoArtist.

Not having a reason to be on the blockchain bothered me because I always felt that the medium should reflect and add to the art. As I alluded to earlier, the medium should have a reason. To demonstrate what I mean by this, one of my favorite pieces of art is a collection of 14 self-portrait busts called Lick and Lather by Janine Antoni. In this piece 7 of the busts are made of chocolate and 7 of soap. There are many interpretations of this work but my favorite is that the chocolate represent desire and the soap beauty, and together it is an exploration of the societal expectation that a woman be both desired and beautiful. The busts could have been marble, bronze, or any of a number of mediums but chocolate and soap were chosen with a purpose. All art should be this deliberate.

For similar reasons, I intentionally have my robots paint with a brush on canvas. There are simpler ways to use paintings robots. I could avoid the mess and just use markers. Or even easier, use a manufactured plotter printer to execute my strokes. But I choose brush on canvas with a robotic arm because I am making a reference to the fact that these machines can have a creative process like human painters. It is also a reference to the long historic significance of painting to fine art. So the medium of a painted canvas is important and meaningful to my art. It elevates the work from being a simple mechanical reproduction and adds context.

So in this regard there is a reason my robots to paint on canvas. Regarding the blockchain, however, I could not find a valid reason to mint the photos of these paintings. One justification for minting digital work is that it creates scarcity, allowing the digital work to hold value. This is a valid reason to mint digital work, but as you can see it did not apply to my robot paintings. The paintings were already 1/1s and minting them did nothing to help with their scarcity. Because of my personal unrest with whether or not I should be minting photos of physical paintings, I started exploring alternatives.

In doing so I stumbled upon my bitGAN series. For those unfamiliar, they were a statement on PFP projects that had become popular in the space. While I question whether most PFP projects are actually art, I do see the potential for them to be, and thus I started using my AI to create bespoke bitGANs. The series grew into a number of projects including podGANs and my most recent byteGANs. My art had not experience such popularity before. A crypto native audience attached themselves to the series and just two years later they have collectively achieved over 8,000 ETH of trading volume.

The most popular of the series has been the byteGANs, both for the fact that they match the pixelated crypto aesthetic, but also for the fact that they are 100% on-chain. Their medium is Ethereum where 1111 11x11 GIFS have been directly etched onto the blockchain. Unlike my robot paintings which were physical pieces on canvas, these byteGANs were digital pieces on the blockchain. Was there a reason to mint them on the blockchain? Of course because they could not exist without it. But more than just their existence, the byteGANs explore multiple elements of cryptoArt, both overtly and with subtlety.

Overtly they have rarity traits which are extremely popular with cryptoArt collectors. Also they make great PFPs. In addition they serve as a social token both for supporting me as the artist, but also indicating that the collectors are in a club that thinks 100% on-chain art is valuable. More subtly, these have been called “dancing coins” and are treated as such. I find it interesting that throughout history, civilization has always put art on their currency. The byteGANs are an artistic currency. But even more interestingly they balance right on the edge of being an art project and an alt-coin. I would estimate that half of the collectors considers byteGANs an art project, and the other half considers it a speculative commodity. All of this comes together to make their use of the blockchain not only purposeful, but essential to their meaning.

My fascination with all of these facets of the blockchain spilled over into my other AI art. I began to wonder if my more complicated projects could be put on-chain. Collectors had long been asking for more AI Imagined Faces, even offering up to 10 ETH for the images I recorded on BTC with Verisart. But I had always declined because there was no reason to return to the project. But then, quite coincidentally, BrightMoments asked me if it was possible to put 100 AI Art pieces on-chain for their Tokyo exhibition. Since I was already thinking about AI Imagined Faces, I began to exploring if I could. There was one major challenge, however. I wasn’t sure I could afford to because of their file size.

Recording art on-chain is prohibitively expensive and a function of the filesize. For example, recording the 11x11 pixel byteGANs on-chain cost 30 ETH (Note: while some solidity devs claimed it could have been done for less at the time of release, recent reviews with their suggestion revealed a maximum of 10-20% savings). Recording high resolution AI animations would cost considerably more on the order of about an 0.1 ETH for every 30KB. At todays rates that is about $6,000 USD per MB and the animation were approximately 5 MB. So the challenge was taking my old AI model from 2018, making 100 new faces and compressing the 5 MB down to a size that looked good, but was also inexpensive enough to be inscribed them on the blockchain.

Using all the digital tricks I knew, as well as inventing some new ones, I generated 100 new AI Imagined Faces and converted their animations to line art. I was then able to further compress the line art to under 30KB and inscribe each on the blockchain. The whole process used multiple layers and techniques of compression and ended up costing approximately 10 ETH. Could it have been less? Probably, but not by much. Also even if I could reduce the cost, I would have used the savings to improve the quality of the animations. !0 ETH was what I wanted to spend on preserving this art in as high a quality as possible. I am please with the results. Considering how these are line art, they could scale up to the size of a billboard without a loss of quality,

Was minting 100 artworks for this project worth it 10 ETH? Absolutely. AI Imagined Faces : On-Chain is a fitting end to the series and this chapter of my artistic journey. Five years ago I was an AI Artist who was only using the blockchain to make more sales. Today I am a cryptoArtist that is using blockchain to do things with my art that is not otherwise possible. With my 100% on-chain work, the blockchain has quite literally become my canvas. Looking forward to what the next 5 years hold.

Pindar Van Arman