Micah Elizabeth Scott won the Web Workers Dev Derby with Zen photon garden, her impressive (and fun) interactive web raytracer. Recently, I had the chance to learn more about Micah: her work, her ambitions, and her thoughts on the future of web development.
How did you become interested in web development?
I’ve been into building things for as long as I can remember. I love
making things, and I’ll often learn new tools just for the sake of
giving myself a different set of challenges and constraints to work
with. My first big web project was an early collaboration tool for
open source development, dubbed “CIA” because it spies on your source
Can you tell us a little about how Zen photon garden works?
Zen photon garden is a type of raytracer, which is to say it simulates
the path that individual rays of light take as they bounce around in a
scene. It’s a two-dimensional raytracer though, which opens up kind of
a neat new possibility for visualizing how light works.
A traditional three-dimensional raytracer traces rays “backwards”,
casting rays out from each pixel on a virtual camera, bouncing it off
of the objects in your scene, until it finally reaches a source of
light. Each pixel of the scene comes about by counting, on average,
how many photons would reach that portion of the virtual camera.
In Zen photon garden, light rays emanate from a lamp and move along
the image plane in two dimensions. Instead of visualizing the single
point where a ray reaches the camera, I visualize the entire ray as it
bounces through the scene. Each ray turns into a sequence of line
segments, beginning at the light source and bouncing off of any number
of objects before it’s eventually absorbed. This process repeats
hundreds of thousands of times, and the image you see is a statistical
average of these many light rays.
The inner loop of Zen photon garden is quite specialized. For each
light ray, I need to trace its path by intersecting it with the
objects in the scene, and each segment of this path is visualized by
drawing an anti-aliased line into a high-dynamic-range 32-bit
accumulation buffer. After tracing a bunch of these rays, the
high-dynamic-range buffer is mapped to an 8-bit-per-channel image
according to the current camera exposure setting, and that image is
drawn to a Canvas.
These anti-aliased lines need to be fast and very high quality. Any
errors in the uniformity of the line’s brightness, for example, will
affect the smoothness of the final image. To get the combination of
speed and accuracy I need, this line drawing algorithm is implemented
be managed carefully so that the app can draw with high throughput
when you leave it alone, but it can still respond with low latency
when you’re interactively adding objects to the scene.
What was your biggest challenge in developing Zen photon garden?
The hardest part of implementing Zen photon garden was making it run
as fast as possible on all of the latest web browsers. Thankfully
these days it’s relatively easy to write an app that runs on all
browsers, but making it run optimally is tricky when your application
is CPU-bound. Small changes to the inner loops would cause big
This required a lot of trial and error, and a few trips back to the
What makes the web an exciting platform for you?
To me the killer feature of the web is its universality. Modern web
browsers are nearly ubiquitous, and it’s the fastest way to take a
weird new experimental concept and get it into people’s hands right
now. As someone who loves exploring the intersection of art and
technology, this means it’s finally possible to send your friends a
link to your latest art project without having to worry about what
operating system they’re using or whether they have the right library
What new web technologies are you most excited about?
WebGL is really exciting to me, but as someone who used to write
graphics drivers and worry about security for a living it also kind of
The web technology I’m most excited about would have to be asm.js
actually. I’ve always enjoyed getting my hands dirty with low-level
graphics code, and even in today’s world of GPU acceleration and
high-level 2D canvas APIs, I still find plenty of reasons to push
pixels. Having a way to get near-native performance in a very reliable
way across all major browsers would open up some great new creative
possibilities, and I’m excited to see where that leads.
If you could change one thing about the web, what would it be?
It’d be great if we could find a way to ease the tension between those
who see the web as a content platform and those who see it as a
software operating system. Right now it feels like HTML is too
unwieldy to be a document markup language, and it’s just barely
starting to get the services you’d expect from a modern operating
Do you have any advice for other ambitious web developers?
Plan to prototype a lot of things, keep the ideas that stick, and
throw the rest away. Respect the web as a platform, and try to be
playful about exploring its margins. Understand but don’t begrudge the
ways in which web programming is different from other kinds of