If the overhead of a subversion checkout storing each file twice is too much for your dataset and drive, you might take a look using scord with subversion, which allows your checkout to store only one copy of each unmodified file. http://scord.sourceforge.net/

A little more detail: In addition to the working copy of a file, a subversion checkout includes a pristine version of each file for diffing (e.g., 'svn diff', as well as to make 'svn update' and 'svn commit' more efficient). scord is a Linux and Mac OS X FUSE file system that mediates access to a directory tree, detects when a pristine and working copy pair contain the same content, and keeps only one copy around in these cases.

I store my photo album in subversion - I love not having to worry about accidentally losing a photo. Halving the disk space needed for the checkout lets me fit my photo album into my laptop harddrive.

Disclaimer: I'm the primary author of scord.