More technical information is here:
http://www.fuel-3d.com/files/Fuel3D_Whitepaper_1.0.pdf [!! .PDF !!]
Even more technical info - it has equations and everything:
http://carlos-hernandez.org/papers/icvss_chapter2010_photometric.pdf
There is software available that can create - with some human intervention - 3D models from photographs, but it relies on the cameras being calibrated (lens aberrations will throw it off the scent) and work better for obkects that aren't moving. Moving objects can be captured, but that requires several cameras to synchronised. The cheapest way to do that is probably with some low-end Canon compact cameras, since many models support the use of a temporary firmware, CHDK, which allows home-made remote shutter releases to be used.
AutoDesk also have a service where you can upload a series of 2D photographs, and receive back a 3D model.
Personally, I'm tempted to wait and see how well Intel's upcoming 'RealSense' scanner performs - it's likely to be far cheaper than this, though aimed at a different market.
Ideally, I'd like a 3D scanner to be able to give positional feedback to a 3D printer during the printing process - i.e automate the axis calibration process and correct for any errors that occur.
For 'scanning' shiny things (machined metal) to a very high level of accuracy, you want want of these probes from the British company Renishaw (oh, and an expensive CNC machine to mount it on):
http://www.youtube.com/watch?v=JPFB7NiW0UY
Another of their probes was featured 4:57 the iPhone 5 video, without Renishaw knowing about it - it's rare for Apple to featured a branded object that isn't theirs in their marketing.
(The fancy house featured in the latest episode of Sherlock was built by Renishaw's founder)