Building a place for the birds to put their foots.
http://literal-trees.co
https://ahw.github.io/literal-trees
- Everything starts with Math.random. The seed-random library provides a PRNG that overrides Math.random so you can get reproducible sequences of psuedo-random numbers given the same input seed. That's how the persistant link works.
- A lot of the random values are coming from the Normal distribution. I'm pulling values from that distribution using the box-muller module.
- The tree-rendering code is actually kind of basic — it's just a BFS algorithm. Still, it takes a fair amount of time and processing power. To unblock the main thread it's all running in a WebWorker. The WebWorker runs the BFS algorithm and returns a giant stringified version of the SVG document. In an earlier version I was building the SVG in the DOM directly using RaphaelJS and it was 60X slower.
- Back in the main thread, the SVG document is rasterized client-side (!) into image/png data using HTML5 canvas. Data URLs abound, as do onload callbacks.
- If the device screen is taller than it is wide I assume you're on a phone and pad the image as necessary so that it matches the aspect ratio of the device exactly. This makes it easy to just save the image and have it work seamlessly as a phone background or whatever.
- Print styles! I spent some time (a lot) fiddling with the print styles. If the height or width of the tree doesn't fit on a single page, then the values of maxprintheight and maxprintwidth come into play in order to scale the image.
- Works offline using ApplicationCache. Why? Just 'cause.
My name is Andrew Hallagan. I enjoy this sort of thing and I hope you do too. This is my first foray into generative art; I'd love to hear what you think about this or anything similar via email or Twitter, where I am @andrewhallagan.