sangsara.net

(Edited)

One of the most amazing photo processing apps I’ve ever seen. Takes the premise of intelligent tools like AutoStitch and amps them a thousandfold. It’s being done at Microsoft now, but I think the presenter said that the technology was acquired along with a company.

Essentially, PhotoSynth grabs all the pictures of a given area, and recreates the 3D space by comparing data. The dataset shown in the video is Notre Dame cathedral made entirely from tourist photos found on flickr. Using the power of crowds to document something, the power of the web to aggregate the data, and technology to assemble them. In the process it creates a richer metadata context, automatically determining relationships.

Edit: Turns out it was announced last year, and small video demos have been available for awhile now. The video above came from this year’s TED conference in March. I found it via Slashdot today, and the BBC has plans to use the technology in a new documentary on How We Built Britain.

Of equal interest is the SeaDragon image scaling technology behind PhotoSynth. You can see it in action before the PhotoSynth demo in the video above. Straight out of an SF movie!

This is Microsoft’s corporate video of PhotoSynth. Also demos what it’s capable of: