3pointD.com

This landed while I was on hiatus, but is so cool and 3pointD that I can’t help but blog it here. Microsoft Live Labs’ Photosynth technology (in preview) matches up hundreds of photographs of the same location to create an image space that’s navigable in three dimensions. The technology works by making a “point map” of each image and then mapping all the images together in the same space. Besides the fact that you can make a 3D space out of photos that were casually taken by a hundred different tourists, say, what’s cool about this is that the software retains metadata for each photo, so that you can drill into an image and then pop out the other side at the Web page of the person who took the picture.

There are a couple of cool videos to check out, including this short one and a longer one on the Microsoft site.

One cool thing about this is that it relies heavily on the kind of accidental collaboration of Web 2.0. From the Live Labs Web site:

Your brain knows that your eyes are about two inches apart. But when Photosynth does its magic, it doesn’t know where the cameras were, or which way they were pointing. Fortunately, when there are many cameras, and many features in common, the algorithms behind Photosynth can figure out not only where the features are in 3D, but where all of the cameras would have to have been, and which way they were aimed, consistent with the features they “saw”.

The software essentially needs a critical mass of data, which allows it to draw conclusions about the position of each photo relative to every other in the batch. The more people snapping the same general location, the better the data. As nailchipper points out, “What’s exciting about this technology is that one day it might be possible to run this with a large photo repository, such as flickr, where there are millions of photos of places all over the world.”

Imagine that: a system that automatically matched up every photo on Flickr to create a 3D-navigable model of the entire world — or at least, as much of it as had been sufficiently photographed. And because the software creates a point map from the architectural and other features in each photo, it should be easily possible to use it to create 3D models that could be imported into a shared virtual space, as well.

I love the fact that you can push through the “place” and into a related Web site as well. This is the power of such mirror worlds, as I see them: not necessarily to mirror the world per se, but to add information and destinations on top of the re-created image. This is the new kind of connectivity that I think is going to make 3pointD applications so powerful and useful. It’s one thing to make a virtual Venice out of a bunch of snapshots, quite another to let each of those snapshots lead you where they will, whether into an online museum, an architecture site, a preservation society, or just Aunt Tilly’s blog of her Italian vacation. Nice stuff.

[Scott McMillin originally tipped me off to this, but it also comes via Very Spatial.]