Around the edges of WWDC: techno-porn

Wednesday through Friday they've had lunch-time talks as well (so, obviously I've only been to two so far).  These are general interest subjects, usually with a vaguely mac-related theme.  Last time I was here they had a guy from Pixar talking about their tool development (a lot more interesting than it sounds :)) and how they put movies together, which was great.

There was a pretty interesting one on Wednesday by the guy from "Smule", an iphone app company.  He's actually a Stanford PhD who did his thesis on a sound generation language called ChucK, which I've come across before but never looked at.  He started with a quick demonstration of the language and environment anyway, doing a live-coding performance :)  Then he moved on to talking about a few "laptop orchestra" projects and the like that had spun off from this work.  Then, he decided to leave an associate professor position at Stanford and form Smule, which is more or less about making music apps for the iphone (as in, apps that turn the iphone into an instrument; although their first was actually a "virtual lighter").  All of them had a social aspect, such as being able to see who's playing what and where, voting on people's performances, etc, and a lot of the talk was about how they try and foster that, and the communities that had sprung up around it.  Fairly interesting, although I won't go into detail.

Last night I also went to a "Coca Heads" session at the Apple store near-by, which was about 6 short talks from developers and designers.  The common theme was definitely attention to detail in design and user-experience, but I won't go into details.  The speaker list is here for anyone who is curious: http://theocacao.com/document.page/606  It was great, I don't expect many of you to be interested though :)

Today's (Thursday) talk though was absolutely sensational!  It was on "techno-archeology", and it covered a lot but the main theme was recovering images from the lunar surveyor films of the 60s.  Prior to the first moon landing they sent up about 5 of these robots to film the surface of the moon, looking for suitable landing places -- this obviously requires a high degree of precision; 1m in this case (since that's about the size of boulder they wanted to avoid landing on).  The camera used had a staggering 24 inch aperture, and of course it was all analogue: the film was scanned on-board in space, then transmitted to 3 locations (including Woomoora) where it was stored.  Here it gets interesting, and I forget the precise details, but their processing "workflow" included photographing it, splicing things together manually by laying them out on the ground, and photographing them again!  (Then they'd manually walk around looking over them for a good landing spot).  Anyway, the tapes survived a couple of close-calls ("should we just ditch them?" "Nah, better hang on to them"), and were actually preserved in optimal conditions.  Next interesting fact: they hadn't degraded at all because, apart from the storage conditions, the binding agent was.... whale oil!  Ooops.  So the data-recovery aspect was in fact all about the hardware: there's only 4 machines left in the world that can read them, and most crucially, only one person with experience on the machines left.  (NASA do have detailed technical documents archived on the transmission format, etc).  It took them 3 months of restoration to get to the point where they could read anything, but the results are simply staggering (he also recounted with great glee about the various letters from people who didn't want to fund the project, stating that they wouldn't see any improvement over what they already had!).

This is the "image of the century":

And this is their restored image (it looks even better on a huge screen, as you can imagine):

The project (still on-going) is documented here: http://moonviews.com/

(Also, the computer that powered those robots had I think 256K of RAM, and performance measured in FLOPs!)

Mark Hepburn 11 June 2009
blog comments powered by Disqus