Generative Art and NFTs

Wikipedia kindly explains that "Generative Art" is a term describing art that has been created, at least in part, through the use of an autonomous system such as a computer program.

To an artist with no technical knowledge such computer programs may seem mystical - how can a non-sentient thing create so much art? And is it even art?

To a dogmatic engineer with little appreciation of the artistic world it may seem equally mysterious - why would anyone think that generative art is special in any way or want to spend time producing it?

Cut it up

When David Bowie claimed he wrote the lyrics for his songs on Lodger using William S. Burrough's cut up method, namely taking written text on paper, cutting it into strips, and randomly rearranging the strips, he was working in a bygone age (i.e. 1979) where this involved serious work. Scissors, glue, a breeze-free room, and a weekend of spare time were required to rearrange and fix down the phrases and sentences.

With a computer program you can achieve the same effect with a few lines of code produced in an afternoon by a half-decent programmer, and what's more, the program can subsequently "cut up" and rearrange the entire text of the Bible and Proust's "à La Recherche Du Temps Perdu" in a few seconds.

Dear artists reading this - seriously, arbitrary "cut up" techniques in the computer science world are considered trivial, even if they may seem magical to you. They have been used in computer games since the 1980s. Every second since computer games first appeared, screens have been showing unique images that have never been seen before, with significance and meaning. Even if that meaning was only: "My laser cannon has been destroyed by fire from an alien".

Dear programmers reading this - some of you may not understand (or even want to understand) why some people get irrationally excited by art. But the fact remains that they do, and some other people out there are happy to spend money on buying it. And just as hand-ground pigments mixed with linseed oil were slowly replaced with factory generated tubes of paint, the art world has been moving towards a greater acceptance that a computer program is as valid a tool as a paintbrush for creating art. And from their perspective, you are in the business of manufacturing new paintbrushes.

NFTs and generative art

Generative art has, generally speaking, been a niche activity. But with the emergence of art NFTs, it has had to step into the limelight. If you are going to release a collection of ten thousand punks, zombies, or monkeys, it is going to take a long time to produce them manually, one by one. Although I'm sure there are at least ten artists out there who have done precisely that in an attempt to emulate a medieval monk's approach to producing such a collection. But for most artists, that's not going to fly. They want to get their work out there in a matter of weeks or months, not years or decades.

And that's where generative art can step in. The principle is very similar to that childhood toy, the paper doll with paper clothes. If you were the niece of a cheapskate uncle back in the the eighties, you might have received one: a collection of card pages with a figure (usually female, in her underwear), and a bunch of dresses and hats and so on, held onto the dolls by paper folding tabs.

The particular kind of generative art that is proving popular with profile picture NFTs (pfps) works on the same principle, and relies on the fact that the png image file format allows for transparent pixels. This means that an artist can, as an illustrative example, design five different base ape images, ten coats or shirts, ten ear ornaments, ten hats, ten eyes, and ten mouths, and then combinatorics means that these templates allow for a total of 500,000 different and unique ape images. So by drawing 55 different images, scanning them, removing the pixels surrounding the relevant pieces of the images, and then using a computer program to overlay and combine one of each of the body, coats, earrings, hats, eyes and mouths, a total of half a million unique images can be created.

The drawing part may take a week or even a month if quality work is being drafted, but after that, producing the images takes less than a few minutes once the computer program to do so has been written.

So the key here is that the amount of work required can be calculated using addition (5+10+10+10+10+10 = 55) but the resulting output is determined by multiplication (5*10*10*10*10*10 = 500,000).

Of course he has to mention the Orthoverse

In the Orthoverse, combinatorics turned out to be rather important. I initially wanted to make sure that each and every token had a unique image and name, which turns out to be a tall order. The tokens themselves are definitely unique, because the token identifier is the same as the address that it originally belongs to. There are about 1.5*10^48 Ethereum addresses (the actual number is 16^40, or 16*16*...*16*16 a total of 40 times).

And the actual images?

Each Orthoverse land consists of a 6 by 6 grid of different geographical features - two different types of plain, a mountain, a hill, a deciduous forest, a spruce forest, a pit, or a lake, so 8 in total. The name for the land, which is shown in the token image, is determined from a total of four syllables, and the number of choices for each syllable are 9, 72, 72, 72 and 40. And the background is selected from a palette of 8 colors. Which of each of these choices to pick is determined by the digits in the token identifier. So that means there are a total of 6*6*8*9*72*72*72*40*8 different token images that can be generated. That's only about 310 billion different unique images. So if someone had the time, they would find plenty of addresses with exactly the same image. In fact, for every image, there are about 5 undecillion tokens with that image.

But in practice, no one is going to do that. There are only about 8 billion people on the planet, very few of them have an Ethereum address, less are interested in NFTs, and even less have heard of the Orthoverse.

Finally, a minuscule number those will complain about the fact that their token looks and is named the same as someone else's.

And in any case, I can just ignore them.

Summary

Generative art involves using a computer program and combinatorics, and possible random numbers, to produce artwork.

From the artist's perspective, this means that they can do a small amount of work, and produce a large number of images. They can then either push the whole lot out there, or go through each one to see if something speaks to them and might therefore be worth publishing.

From the software developer's perspective, this is all a rather trivial activity, involving a few for loops and something like the Jimp package to do the heavy lifting.

Artist - meet developer. Develop - meet artist.

Now both of you: go and have some fun, and learn something about a world that is different to the one you normally live in.

PS: if you want to see what Orthoverse tokens look like, you can reveal your own at the Orthoverse website, or see them on OpenSea or LooksRare.