This technology from Pravin Bhat over at the University of Washington is pretty impressive. It uses high resolution still images to enhance video footage. It also enables dynamic masking and all sorts of other goodness. The interactive cutout video is interesting too - unlike most compositing ways of thinking, which are pretty much working in 2D on a frame-by-frame basis, their system looks through the pixels temporally and allows you to select through them by time in a kind of time smear fashion.
It should make a big impact in post-production, but I can imagine it might come into all sorts of use for both devious manipulation of video, fixing up crappy YouTube clips as well as crime-scene investigation. And “Spacetime Fusion” is such a great name.
(Via Iain at Crackunit)