Hello! Welcome to the first dev log for my generative art environment project. A bit wordy, sorry. I’m still thinking of a name.
I met my friend Heidi a few days ago for coffee, and I mentioned that I’m working on a project and feeling good that I’m making regular, visible (to me) progress, and she mentioned that I should keep a log of the progress I’m making. I gave it a brief thought at the time, but it wasn’t until the next day when I had made a few noteworthy realizations about functionality that I’d like to eventually build that making the dev log to jot these down could be useful and be another way to see progress being made, in a different medium (especially since the project isn’t out in the open yet). I also got a job rejection the same day, and the idea that it could be nice to use the dev logs as a showcase to potential employers crossed my mind as well. So here we are!
I don’t aspire for this to be a weekly dev log — I don’t want to set that kind of “pressure” on this — but I do plan to write regular updates as I have them. I’m going to use this first log as a space to give some background on the project, so it’s not going to be particularly technical. Future logs will dive much more deeply into the “technical weeds”.
A bit of background. I’ve been interested in generative art for a while (since at least 2018), and every once in a while I would get a small spark in my mind and look up again some interesting ideas and think, I’ll get started with making some sketches! As with many things, these kinds of ideas are competing with many other ideas and general things going on in life, and each time I got that spark it would just as quickly fade away until the next time.
A few years ago (I forget when exactly, but either in 2021 or 2022) I got another one of those sparks, and while doing some research I stumbled upon Tyler Hobbs, who is a software developer-turned-generative artist, and I found his works amazing. He makes beautiful works with just algorithms, often with a bit of randomness.
While he doesn’t write as much about his process anymore, he wrote many blog posts detailing some of that process, and as I read a couple of them I noticed, “Hey! He’s using Clojure!” At the time I was working at Pitch and used Clojure daily, and I thought it was cool that he was using the same relatively niche language that I was.
Fast forward to the beginning of 2023 and I begin work on this project. Although I was feeling very inspired by Tyler’s work and technical approach, in particular his use of Clojure and quil (a Processing/p5 wrapper in Clojure), I wanted to approach things from a different direction. From what I could tell, Tyler used quil solely as a Processing wrapper, and his iteration loop looked pretty slow: “write code, output the result to an image/images, tweak, repeat.” I really wanted to be able to write some code and immediately see the changes to my art, rather than outputting to an image first, and since quil also supports p5 (JS version of Processing) and I’m more comfortable with JS in general, I opted to go in the direction of making a web app which would update your image as you write.
I worked on it on and off throughout 2023, eventually culminating in a really rough prototype that worked as expected: I was able to type something on the left, and see it on the right.
I made some nice strides on a technical level along the way too:
- I’m using the ClojureScript compiler directly. Although I respect the immense work that Thomas Heller has done with shadow-cljs, in particular making it much easier to work with JS libraries in ClojureScript, I find the tool quirky, quite heavy, and cumbersome to work with if there are things you need to customize (I had a lot of experience with this when writing cljest at Pitch) or need to do something a bit less typical like self hosted CLJS, so I made the app work without shadow-cljs and instead opted for just the vanilla CLJS compiler. It hasn’t been without its frustrations — the documentation about getting started is very lackluster and I had to do a lot of trial-and-error investigation, and shadow-cljs plasters over and/or fixes bugs that the vanilla compiler suffers from and adds functionality that the vanilla compiler doesn’t have, some of which I’ve also had to manually debug and/or add — but doing it this way forced me to understand certain functionality, like how namespaces are loaded in the self-hosted compiler, that I never would have otherwise.
- Above you can see there are two numbers, the seed and the iteration count respectively. Based on the idea that I would be making static sketches, I wanted to have a bit more precise control over how many times to run the draw loop and how the random and noise seeds were set, so I added functionality allowing just that: you can specify how many times to run the code on the left, and what seed will be used when getting a random number or Perlin noise value from p5.
I kind of abandoned the project throughout 2024. I was focused on some other creative projects, and in particular I spent a lot of creative energy on photography, making a zine and trying out tintype. I also changed to a new job that pretty heavily stressed and ultimately burned me out, and I just didn’t have the mental energy to focus on many other things.
At the end of 2024 I became unemployed, and after a period of decompressing, I start to explore a bit and a spark again comes. I was watching a Coding Train video — which, as an aside, if you’re at all interested in generative art and/or artistic programming in general, I highly recommend checking out the videos. Dan has an infectious energy and there’s a lot ot take away from each of them. Anyway the video in question featured an artist named Patt Vira, who has her own channel of coding tutorials, and I eventually watch her video about “string art”. While watching I thought that 1. since I have draw loop iterations set up in my environment, I could see the algorithm work step-by-step, and 2. it would be cool to try this on some of my own photos, and while I haven’t actually wrote the sketch to do the string art, this spark ultimately pushed me to begin working on the project again.
By today, a month and change later, I’ve not only revived the project but made a lot of progress:
- I’m now rendering the sketches in an OffscreenCanvas and in a web worker, which helps keep the UI working smoothly when the sketch is computationally expensive, and can be used to put a sketch in the background of various pages, like the login page.
- I’ve started using p5.js 2.0, which while still in beta has a lot of improvements over version 1, such as allowing the use of the OKLCH color space. This, among other reasons, necessitated moving away from quil and so now I have a custom wrapper for p5.
- I’ve improved the way that libraries are loaded for the self-hosted compiler, so it’s easier to customize how they’re loaded and registered in the browser.
And I have some ideas for the future:
- What would it look like if the concepts around “setup” and “draw” were made generic, and you could extend your sketch to have more than a single draw loop?
- Creating art is an iterative process, and it would be nice to be able to see the history of that process, rather than solely the current state. It would be just as cool to be able to look at that history and branch off from it, akin to git.
- Speaking of iteration, sometimes the initial idea you had doesn’t perform well and you want to optimize it to run faster. Are there intuitive ways to help someone see slow parts of their code?
- Right now it only supports p5.js, but I’m directly creating and passing a canvas around. What if it could support other libraries dynamically, like d3 and GSAP?
- What if you could define discrete steps and transition points between those steps? Could you make a programmable slide show? This is the idea that pushed me to write down a dev log.
- What’s the fun if I’m the only one able to use it? I want to deploy it to my server, so that others can too.
It’s still very rough - hell, it doesn’t even let you save a sketch right now - but I’m excited for the future.◾️