The Mechanical Turk

Our congratulations go out to Ian White, Argument from Design’s lead programmer, who got married to Tessa Mae Peasgood on Wednesday! We wish them the absolute best.

Raja Sandhu has written a FAQ regarding the development of logo — or, more accurately, corporate identity — design.

It’s good to see someone else taking a rational approach to design — or, at least, as rational approach as one can take. Many, if not all of Raja’s comments reflect our own experience of corporate identity design. In fact Raja’s comments can be expanded to cover graphic design in general. A good read!

So Ian’s been extolling the virtues of the Mac text editor TextMate for some time now. I’ve been dubious, because I’m a long-time BBEdit user — I’m a big fan of BBEdit’s HTML palette. But in a brief session, Ian showed me how easy it was to added “snippets” of code for long-standing quirks: for instance, “curly quotes” and other typographic titbits.

So he sold me on the application, and I’m making a switch to TextMate. The creation of bundles of typographic and/or frequently-used (X)HTML will make my life a lot easier. If you’re a TextMate user, you may be interested in the bundles we’ve got available online. They may be very helpful.

For fear that they''ll look like this:

A little note on why we call our blog the “Mechanical Turk”. Pace Amazon's Mechanical Turk, we were drawn to the idea of something that appears to be artificial intelligence, but in reality is just a performance of human ingenuity. It speaks to what our web design & development is all about — work that looks very slick & polished but is the result of of craft.

You can read more about the actual Mechanical Turk on Wikipedia.

Now that we've got our shiny new server, I need to find the time to write about, and better document, the 30 or so rails plugins that we've developed over the last few months.

When writing library code for projects lately I've been using the 'create a plugin' mentality instead of 'create my own framework'. The former is better because (i) the less dependencies a piece of code has the more likely it will get reused, so (ii) it will get better and stay up to date with the underlying framework, and (iii) a small independent piece of functionality is easier to define, understand, and write tests for.

So stay tuned for some write ups, starting with inherit_views; a plugin that lets your controllers inherit views from other controllers.

One of Argument from Design's specialities is the creation of virtual reality panoramas. But the major problem in producing virtual reality has been the problem of creating, or "stitching" photographs together: while theoretically you could take just two 180° fisheye images & stitch them together, you in fact couldn't, because a company called Ipix in the US held the patent for stitching together fisheye images — and they didn't license their patent to any other companies. You could only legally stitch together VR panoramas if you used their own software, which was frankly terrible — the results were awful & they insisted that they owned the rights to your images. Not a good way to do business. Ipix was known as "Ipox" in the vr community for these reasons.

So we all had to limp along with tedious & ridiculous workarounds to produce our panoramas from fisheye lenses — we'd have to take six fisheye images, convert them into wide-angle images (and thereby lose lots of information per image) and stitch them together. This was frustrating (to say the least) because the more photographs you have to take, the more work you'll have stitching them together — clouds move & change ambient light, and correcting these effects takes up production time. And that costs our clients more.

Well, it looks like Ipix won't be a problem for any of us any longer — they just filed for bankruptcy. Immediately, Realviz (makers of Stitcher) shipped an update that helps us make VR panoramas directly from — you guessed it — fisheye lenses. This is a boon for us, but even more so it's a boon for our clients: the barrier-to-entry — the price of production — will now go way down.