Robot Dreams

A few days ago I posted a panorama of the Toronto skyline. It turns out it was done by hand by loading up the images, doing some blending between them, aligning them, and brightness+contrast adjustments. And today, while trying to do the same to another set of images I failed. And I thought maybe someone has written some software to make such things easy. Well they have, somewhat. It turns out panorama stitching is not exactly easy and just about all the solutions out there are manual in more ways than one. Except one tool, which comes from the AI lab at the University of British Columbia, Autostitch. So I decided to run it through some of the multi-shot panorama images I have, really only three sets since I'm not a panorama fan. Below are two of the results. The first from the same set of images as that previous Toronto skyline post (warning this image is even larger than before), and the second from an earlier set of phone images taken at Kauai. But it figures that the set of two images that started this endeavor, Autostitch failed to put together. Neither man nor AI could solve the problem, maybe I just need to wait for some extraterrestrial intelligence.

Right before I left on vacation for Kauai I replaced my phone service. There came a point that I could not receive phone calls while sitting at home, and I've had only a cell phone for years now. To put it mildly Cingular's claims about network coverage doesn't cover the customers that got swallowed up by their ATT wireless purchase. So I switched to T-mobile, for a lower price, more minutes, new international phone, and now I get a full signal at home

Why I came to use my phone for some pictures in this trip is simple... I didn't have time, I'm the kind of person who packs in the last 5 minutes if possible, to test if my SLR camera was working. So I try and use it for the first time in the helicopter tour we took, and it doesn't work. It just decides to rewind every roll of film I put in it. Anyway here are some of the resulting phone pictures from the trip...

• We tried doing a small hike to a pair of waterfalls, we only managed to see one of them after getting lost, and along the way we had some interesting spectators.

• One of the first things we did was to scoped out some of the local beaches and parks. The first one we visited was Ke'e Beach at the end of the road entering the Ha'ena State Park. None of us that went are really beach persons, John and I where more interested in the snorkeling, of which I'll post pictures eventually.

• The town we stayed in, Princeville, was once part of a plantation as much of Kaua'i was. At this point it is mostly resorts and vacation homes. But across the Kuhio highway, an overrated term since it's only the speed limit not the size that seems to change, is a lookout point with a view of the Taro Farms and of the start of the mountain range behind it.

• And stragely, on the last day, we took a stroll down to the beach at the end of the street in Princeville. The Puu Poa Beach is at the bottom of the cliff on which stands the Princeville Hotel. The image in this case is an arrangement of three phone pictures taken from the same spot.
  

Shortly before I went on vacation to Kaua'i North Shore, Princeville Google started it's second Summer of Code foray. Like usual I can't resist making suggestions and put together a very short page in the Boost Wiki with two project ideas, the first two Boost.Build ones. Like the proverbial snowball rolling down, many other people added ideas. And at some point, one day past the deadline, Boost was accepted into the SoC. And right before I left for Hawaii, I joined the mentor list for Boost SoC projects. So here I am in vacation, having fun, except for the part of loosing my wallet at the beach, and the Boost SoC wheels keep grinding.

So I come back a week later, one week ago now, and a few thousand emails of all kinds have accumulated (yes I got more than 3000 emails in one week). After quickly parsing a small percentage of the email, I go look at how the Boost SoC applications from students and find 130 still active applications to go through (44 of them had already been tossed out by the other mentors). Yikes I would have never expected so many students wanting to do work for Boost. As it turns out Boost is one of the top projects in the SoC, in terms of number of submissions. So here I am, late at night evaluating more applications. In this case for 2D Geometry Computation, and I've already gone through and evaluated the Boost.Build related ones (turns out there where a small number of those), and the ones for a Generic Tree Container.

I have my own theories as to why students would gravitate to wanting to work on Boost libraries. But I'd like to hear from others, since I haven't been a student for a long time now. What drives students to Boost and C++?

PS. I'll get around to posting the pictures, mostly underwater ones, from the vacation soon.