I was amazed when I first saw this as it combined two main technologies, one being their Seadragon image zooming and the other a technology that literally analyses and stitches together images in to a 3D space (no not the Quicktime equivalent).

For me though the future link is when they line up the idea of dynamically pulling semantic content from Flickr, because the content is already tagged, its a no-brainer to see how humanity could document the world through images.

The exciting part is linking this back to broader content through the existing semantic connections already created.

You maybe already interested in viewing the image you have in front of you, but at a right click you could view related content (Wikipedia?), or even take a look at the profile of the person that took the picture in the first place.

You can also see the connection of 'Multi Touch Screens' being able to navigate using your fingers would be more intuitive.