There was a scene from Minority Report that showed a UI that allowed a person to explore photos and videos of a particular space and time in a fluid way. After watching the scene a few times I think it will be a powerful tool when it becomes reality. I can imagine the day when powerful home computers can stitch together photos and video scenes of a particular place and combine them with other images found via the Net. It should be some fun indeed.

So I stumbled upon Photosynth from Microsoft Live Labs (You can download Photosynth now btw. . ) and also noticed today that the GeoTagger was slashdotted. GeoTagger looks like a super cool little gadget. If you haven't checked it out.. it mounts on the hot shoe of a camera and records:
- Latitude and Longitude
- Magnetic compass direction
- Time and date
Stick the GeoTagger on your camera and you can call the photos up later in a UI that associates the photo with time and space. If there aren't already cameras out there that record all of this data then they'll be coming soon I'm sure. And if you have a GPS receiver you can just use a software solution (there are a few out there like this command line one) to mash a GPS track with your photos without having to buy an extra gadget.

I'm not entirely sure the software behind photosynth is really necessary? I dunno.. but from what was said in the video it seemed like it can process the photos and place them into a virtual 3D space based on the lines and features in the photos.. therefore stitching together panoramas from multiple photos that are possibly from different people and taken at different times. It could be fun, or could be an over the top gizmo that people don't really need to use. It'll be fun to find out.

As for the UI.. take a look at the Photosynth video and then ponder that Geotagging of photos to particular places has been going on for quite a while via Flickr and Google Earth. The existing UIs aren't anywhere near what is shown in the photosynth video, but maybe such an elaborate UI isn't needed? Maybe we're better off with a less interactive 2D interface (yeah, I was ear deep in VRML years ago too) and maybe not. If we're expected to interface with a 3D geotagged photo browser via something more advanced than a mouse then I hope it isn't a "pointscreen" like this guy is using, but I'd be just fine with the "multi touch" screen that Jeff Han showed at TED.

And the tiny little Luddite in me stops and wonders about the wisdom in putting millions (tens of millions? hundreds of...) of photos in one place all tagged with time, date, location, etc. and building a repository of visual information for the day that the ultra powerful AI technology gains access to the Net. Oh, forget about waiting for some AI to use this data, what happens when CNN gets ahold of it. w00t!