Can someone explain to me the basic theory behind HDR? How
are these images captured etc.? I know they are used a lot in
3D applications and I have a bunch of links about all of this but
I'd really like a layman's explenation on what it is exactly and
how the stills are captured.

No-one?

HDR is an attempt to increase the dynamic range of stills or video. A series of exposures (usually varying the shutter speed) are made and in essence, digitally sandwiched together. In the process of merging the data, the over and under exposed portions of each frame are not used. So, shadow information is used from the over exposed shots and highlight information from the under exposed shots. The Kodak DC290 used a script format to automate the exposure portion.

I've done this (not true HDR) with still images to preserve detail in great portions of the digital image.

I don't know about the HDR method Jeff is talking about but I believe you are talking about the method made popular by Paul Debovec and originally created by, uh, now his name slips my mind.

Originally this was used in the computer graphics field to move the range of luminance created in a scene into a luminance range that can be presented by the medium. Most computer systems can display 256 levels in each range of red, blue and green. However, the true range of lighting in a scene, such as the sun shining into the dark recesses of an underground cave, is much wider than that (10,000 or more).

So, like video, the whites get blown out and you lose all the detail in the shadow. The idea is to take the range of luminance and fit them into these 256 levels so the detail can be seen at both ends.

There are a couple of ways to do this but it involves math and software. I am only aware of how this is done generating 3d graphic images but have not read about this using video.