Syncing a couple of gig of material via git will take up lots of CPU & memory, including on the server, to deal with recompression & history, won't it?