Andy Piper

2024-07-30 17:16:50

I’ve given some talks in the past 6 months about the history of computer art, and in particular art created with pen plotters and drawing machines. As I got into building plotters last year, I didn’t initially think too much about the background to what I was doing; but, being an historian, I then started to dig into it, looking back to the emergence of computer art of the 1950s and 1960s.

My personal favourite piece, Schotter (“Gravel”) by Georg Nees, was made using an algorithm.

At a high level, it’s a simple and very effective rule – draw a square; repeat, adding a small but increasing amount of rotation (noise) with each column and row. Such a basic piece of code produces a wonderful and pleasing – to my eyes – “disintegration” effect. My description is a very simplistic way of understanding the code – more recently, Zellyn Hunter has done a fabulous two part deep dive on the program, going as far as recovering the random seed used to create the specific piece of work that is part of the collection at the V&A Museum in London¹. Until about a month ago I had an alternate / approximated version² hanging just inside the door of the studio, and thanks to Zellyn I now have a precise recreation generated from Python, plotted using my own machine, on fine black paper using gold ink.
Schotter, in the V&A Museum, LondonRecreation of Schotter, Forge & Craft studio, London
In my talks, I’ve joked that we can think of these in-their-time “magical” programs as being like AI, but in old school terms, they are just algorithms.

To say that today’s AI is “just algorithms” would be ridiculously reductive – that form of AI involves not just algorithms, but vast Large Language Models that inform the outcomes of the generated art or text or code; but ultimately it is just super-powered autocorrect and applied statistics³, which derives superpowers from the corpus of data that it has been trained on. I completely understand the concerns around how that data is being acquired and (mis)used, and the skepticism and trepidation and other reasons folks have to disdain “AI”.⁴

One of the common things we get asked when showing our plotter art is, literally: "where is the art in this?" (thus the title and inspiration for my talk); or "so, the computer did it / it's all done with AI then?" (the implication being that we didn’t really make any creative effort of our own).

In our (Forge & Craft case), the work tends to divide between two types and styles (not exclusively, but mostly):

• things I generated using code, mostly experimenting in different languages and frameworks as I learn – this is usually referred to as “generative art”;
• images taken as photographs by one of us, transformed into plottable line art using algorithms (we use DrawingBotV3 for this, it has an amazing range of path finding modules, excellent support, and solid knowledge of a range of different plotters including output to Inkscape SVG for AxiDraw, or to HPGL for my vintage plotter).

Is that “AI”? No – we’re not using what is known as generative AI in those cases. I have, of course, experimented with tools like Midjourney and so forth for some limited image generation, and for some coding assistance – I’m a curious technologist, and I like to explore new tools – but, in terms of putting down the lines in the plots, we’re using algorithms to derive the pen paths.

We also work with analogue materials i.e. the pens and inks and papers are themselves an artistic set of choices. Take a couple of my other pieces, 1984, and Cellular.

The piece on the left is a Cistercian numeral or cipher. I used Python code to generate the glyph here; I used DrawingBotV3 to process the SVG into a plottable format (a close look would reveal it is a dense squiggly line fill); this was plotted using sepia fineliner onto cotton rag paper. The piece on the right was made using an online generation tool, and plotted using bronze Uniball onto the same cotton rag paper.

In both cases I’m showing the dichotomy of ostensibly old formats (handmade paper, Cistercian numerals) and modernity (the notion of “1984” be it the Orwellian, or the Apple ad; the bronze metallic ink); and doing that via the physicality of analogue output. Oh, as a side note, this cotton rag paper is pretty challenging to plot onto – it undulates so the pen can easily drag unexpectedly, and I found that some “babysitting” of the machine was required. No AI there.

In relation to the analog element, I love these words from Freya Marshall in the recent book Tracing the Line:

… to see an image take shape, line by line, is […] hypnotic. In the age of social media, where hour-long plots can be condensed into 20-second videos that are instantly attention-grabbing, it is not difficult to see how the pen plotter has risen in popularity […]

Freya Marshall, Tracing the Line

Another author, Carl Lostritto, writes beautifully about this in Computational Drawing:

… when drawing with ink, time and material are consequential.

Even though the volume of ink is almost never enough to provide measurable depth, the behavior of the material ink as it interacts with the material paper participates in cuing and undermining depth. Ink affects paper and paper affects ink. […]

Ironically, pen-plotters, machines that move a physical pen in two axes across paper, were marketed as output devices for architectural and engineering technical drawings. Now, almost all have been re-appropriated by artists, designers, and architects who cherish them for the very qualities that made them obsolete.

Carl Lostritto, Computational Drawing

---

My blog post today was prompted by being asked whether the generative art stickers I wrote about, created by my friend bleeptrack, use AI. Well, first of all, it’s not my art so I’m not going to go deep into explaining or defending the work itself (there is an explainer and video on the website, and both mention the generative and algorithmic nature of the pieces). Secondly I coincidentally also saw today, an excellent piece by Monokai that seeks to separate algorithmic art from generative art and from AI, for many of the same reasons I’ve done above – it also mentions a number of the original algorithmic artists from the 1960s. I also prefer to specify that I’m talking about computer art in my historical coverage, rather than digital art or net art or demo party art (or or or…), which are distinct genres. By the way, folks like Rev Dan Catt are doing some incredible work with AI: creating a personalised assistant, Kitty; and applying the technology tools and his amazing brain to creating ever more impressive artwork – see his Artist Statement.

I’m not here to bash on “AI”, but I do want to draw some distinctions around how and when and where tools are used, and assumptions are made. I’m no apologist for OpenAI, Anthropic, Perplexity and others; I’m an informed skeptic. A kneejerk reaction of “AI = bad” does not take account of how the term “AI” is being misused everywhere to justify investments and valuations – a lot of stuff we previously labelled “algorithmic” is suddenly a flashy new “AI feature”, when it remains simply a complex set of advanced computing techniques and mathematics; equally, separating out an assumption that all “AI” is the result of misappropriated data in a large language model, is important.

Finally, I don’t have videos from the talks I’ve given (at QCon, and at EMF) available to link to at the moment, but I made a small webpage with more information and links to my sources, if you would like to do more reading around the topics of the history of computer art, pen plotters, and how to get involved. I hope you find that interesting and useful!

---

1. I visited the V&A earlier this year and asked to spend time examining this piece, and several other early computer art pieces, in the reading room – part of my research for the “Where Is The Art?” talk. ↩︎
2. I generated a version using the Whiskers library in Rust, which comes with a handy interactive demo GUI for playing with the same style of piece. ↩︎
3. It is also not Intelligent, just backed by a lot of data that makes it sound clever. ↩︎
4. I am particularly angered by the underhand approaches that various organisations are taking, intentionally subverting and ignoring long-held internet norms and contracts like robots.txt, but that’s a different blog post, for another day. ↩︎

Share this post from your fediverse server
https:// Share

This server does not support sharing. Please visit .

andypiper.co.uk/2024/07/30/art…

#100DaysToOffload #art #Books #computerArt #digitalArt #generativeAi #generativeArt #penPlotter #plotterArt #presentations #Reading #schotter

Websites are Blocking the Wrong AI Scrapers (Because AI Companies Keep Making New Ones)

Hundreds of sites have put old Anthropic scrapers on their blocklist, while leaving a new one unblocked.

^{Jason Koebler (404 Media)}