Compositing multiple takes into a single seamless shot can be hard work, involving green screens, mattes, and very precise camera positioning. But here's a new technique that only requires a shared area in-shot between two takes, with no "locking off" of the camera necessary
It sounds easy: "Our method instead composites shot together by finding optimal spatiotemporal seams using motion-compensated 3D graph cuts through the video volume"… well, actually, it doesn't sound easy, but it certainly does seem to work.
Have a look at the videos on this page and you'll see exactly what we mean.
Here's a PDF describing the work in more detail. This technique is clearly in the early stages but we can see it becoming an important item in the compositing toolbox.