This is going to depend on how well the system is able to represent the difference between two video files, which in turn is going to depend on how the video files are stored.
Most version control systems are able to handle binary files; they vary in how well they handle them. Some probably just give up and store each version in its entirety.
Presumably you're using some compressed format (i.e., not every pixel of every frame is stored explicitly). If you have a video X
, and you make a small change to it to produce video Y
, are X
and Y
going to have long stretches of identical byte sequences, or is the compression scheme going to scramble everything? If the former, any decent binary diff algorithm should be able to find (and not store) the identical sequences; if not, no such algorithm can do so, unless it's specifically aware of the internals of the video format.
You might actually get better results with a format that doesn't compress the data very aggressively, so it leaves something for the comparison algorithm to work on. [EDIT: This is speculation on my part; I have no actual data to back it up, but it seems like a reasonable guess.]
I know this doesn't actually answer the question, but perhaps it can provide a starting point for you or someone else.