I like the theory behind these plenoptic cameras. I came across precursors about 4 years ago. I've been studying the theory off and on since then. While applications such as post-focus are interesting, I'm keen on other applications as well, such as 3D capture/display.
My current day job is writing software for 3D lenticular displays. These displays don't require any glasses. However most of the imagery for such displays tends to be computer generated because they require input in which eight viewpoints are rendered (or otherwise photographed).
But the data in a plenoptic camera capture, can be processed to reconstruct the necessary eight viewpoints. This is because there is an implicit computable depth in such captures.
I believe there aren't any motion picture versions of a plenoptic camera yet because the memory requirements are ginormous.