Friday 20th January, 2012
Parker

Parker is a chubby bear-type animal who does Parkour. The character was made in Blender, rigged and animated, before being exported as a series of still meshes for 3D printing. I’ve just completed the walk cycle (of six poses), and got them printed by Shapeways. The aim is to see how few distinct models it would take to complete a series of stop motion animation movements: walking, running, jumping, etc.

These prints were an initial test, and it’s revealed a few problems with some of the finer details: the claws are too small, and the tiny feet are useless for balancing the models. I’ve glued the models onto bases to test stop motion animation, but these are pretty distracting, and ruin the nice sense of the character interacting directly with the scene.

I’d like these to be very easy to animate with – the end goal is to offer sets for sale on Shapeways so that anyone can create videos with them. It might be possible to make them bigger so they’d be easier to photograph, and hollow to bring down the printing price. Certainly, the model replacement speeds up the animation process. Here’s a really quick test on my kitchen table, swapping out the models without too much accuracy:

Wednesday 18th January, 2012
#TworseCode

I’ve been writing some Ardiuno code to allow me to tweet using a telegraph key (a morse code tapper). The key is, obviously, just a jumped up switch, so was really simple to wire to the Arduino. In fact, the button is the first circuit the tutorials walk you through.

Edit: Seems I wasn’t first. @rebobaydobay tweeted me a similar but much better version of this project that’s hosted here. Theirs launched a week after this post, but we were both behind this guy. Alas.

#TworseCode Telegraph Key connected to an Arduino

I’ve now got the Arduino listening out for switch connections (which it has to debounce, so as not to add lots of extra dots everywhere), converting these connections to dots and dashes, then transforming those into letters.

It’s been a pretty important piece of work: without it the telegraph key, which I bought at a local market, would be almost completely useless. There is something quite nice about twittering by tapping out your code like a woodpecker too.

So now to get it to actually send the Tweet, I’m caught in indecision. I can either send the completed string to Processing, and use a library like Twitter4J to do the actual Twittering. This seems like a shame, because tethering a Telegraph Key to a laptop would almost seem to frustrate the point of the whole thing.

Better would be to get an ethernet shield, and use it to make a stand alone twitterbox, which would plug into a router. Or better still, I could get a cellular shield and set it up with a sim card so you could #TworseCode on the go.

Also, because we’re not (yet) as versed in Morse / Tworse code, it would be handy if the box had some LEDs to show you which letter you’d just completed.

So, still some way to go.

I did not tap this blog post out in morse code. Maybe next time.

Monday 28th November, 2011
Halftone Depth Map with three.js

I’ve been thinking about this for a while. Robert Hodgin’s Hello Cinder shows, amongst other things, how to create a halftone image by scaling particles based on the color value of an image below. I wanted to see what it’d look like to place the particles in 3D space, and use an image’s color value to determine the z-depth, rather than size, of each particle.

Cinder and C++ certainly give you the processing power to work with a lot of particles, but I wanted to make this a quick, online interactive piece. This version is built with three.js and the canvas tag.

Launch halftone depth map experiment to see it in all its 3D glory. This example pulls in recent Flickr images tagged with “moustache”. Launch it, and click a photo thumbnail to change to that image.

I quite like the low-res look that the 28×28 particle grid gives here, although even that’s enough to have the processor struggling.

Getting remote photos loaded onto the canvas was a bit of a headache, so I’ve made a php proxy to pass these through locally. You can see how to do so here on Stack Overflow. My js file is up here.

Wednesday 2nd November, 2011
Pretty Little Triangles

I mentioned in this post that, having sketched an idea out in Processing, I was going to make it into a Scriptographer tool. Which I have now done. It’s not perfect, but it’s fun. It makes pretty-little-triangles. Attractive, but meaningless.

Here’s the Scriptographer script: prettyLittleTriangles.

Load in a source image from the menu and click start. It’ll draw the first triangle for you. Grab the yellow pencil tool from the bottom of your tool bar, and start clicking on the triangle to subdivide at will.

I’ve been testing it with images at 1200px x 800px, which seem to work well.

Here are some pretty little triangles I made while testing the script:

Tuesday 1st November, 2011
HDR Panoramic Coastline

HDR (high dynamic range) images seem to have earned themselves a bit of a bad reputation. This is possibly because when it’s obvious that an image is HDR, it’s usually because the effect has been overdone.

The over-sharp, glowing edges that HDR (and especially fake HDR) can produce are pretty off-putting, but I found out just how useful it can be when photographing the weather as we travelled into Northern BC. This image was a composite of 12 shots: four images merged into a panoramic, each of which is an HDR image made up of three different exposures.

On Flickr, or see it in its rather oversized glory here (15mb).

Monday 31st October, 2011
Subdividing Equilateral Triangles in Processing

With some time to kill on the Ferry (once it got dark), I was casting about for a quick code project I could play around with. I wanted to use photography as the input to generate an image, and came up with this subdividing triangles thing.

Above: The code output. Triangles.

Above: The source image, Horseshoe Bay.

It’s a Processing sketch that draws a triangle, filled with the colour from its centre point. Clicking on a triangle subdivides it into four, each filled with a new colour from its own center. The idea is that each click will selectively increase the resolution of the image. Parts of the image can be made identifiable, albeit in a triangular-sort-of-way, through furious clicking.

The sketch is a bit quirky – there are some rounding errors that stop the lines lining up perfectly, and the recursive hit testing for clicking on a tri doesn’t always work. Rather than fix these bugs, I’m going to move on and rebuild it in Scriptographer: it’d be a nice tool to have in Illustrator.

If you do want to try it out, the sketch can be downloaded here: equilateral_triangles

Update: I’ve completed a quick Scriptographer thing of this. It’s available here.

Tuesday 18th October, 2011
Canvas: circle packing & physics

As an early experiment for my new portfolio site, I built a visualisation of my recent work using mootools and the canvas. It was my first attempt at some mootools code, and I really like its object oriented approach.

A (modified) version of the experiment is up here, and here’s the code.

At the heart of the visualisation is a circle packing algorithm. I’d been looking into build my own system for this, with something force directed to group categories together, but I discovered the excellent moocirclepack by unwieldy studios so used that instead.

I like this interface, I feel like it’s playful and encourages some exploration. But in the end, there are a few too many layers between arriving on the site and seeing my work. Actually showing the work kind of got in the way of the fun, playful physics, and quickly became a fairly convoluted system of overlays. I’ve switched the graph from being the core of the site to an extra piece visualising my skill set.

Saturday 15th October, 2011
Stilted houses: desk toy

I’ve just uploaded my first model to Shapeways for 3D printing. It’s a mini stilted house with a wooden exterior, styled to look like some of Vancouver’s character homes. We lived in something like this when we first arrived in Vancouver. Sadly, it didn’t have stilts. It flooded repeatedly, and has since been demolished. This model therefore represents an improvement on the existing architecture of the city.

At this stage, the print is just a test for scale and fragility. I’d like to make the floors of the building modular, so that a custom property could be printed and stacked.

At a later stage it’d be nice to do the same with container homes. If you could remove the side panels of a mini-container,  a lot of container home setups could be made with duplicate prints of a single module design. Who doesn’t want an idealised container home on their desk? It’d go well with your miniature Eames chair colleciton.

Thursday 13th October, 2011
I like the part when

It’s obligatory to start at least one Tumblr. In particular, these should be about as specific and limiting a category as possible. And, inevitably, they’ll run out of steam fairly quickly.

Here’s mine: ilikethepartwhen.tumblr.com. A collection of reviews from amazon.com containing the words “I like the part when…”: a tried and tested review format. I like the part when the reviewer over explains the bit they like and why without any punctuation.

My favourite review:
“I like the part when Mr Fox chased the chickens out the henhouse and tried to catch them for his family for dinner because they are poor.”
About Fantastic Mr. Fox
Review by A Customer

Monday 10th October, 2011
Oh shit, you found a secret website.

With a couple of spare hours, Darcy and I put our heads together to come up with something meaningless and fun: ohshityoufoundasecretwebsite.com

It’s a secret, with share buttons. It’s received hundreds of views, from over 30 countries, which we’re obviously very upset about. Because you’re not supposed to tell. It’s a secret.