I tried both and rejected both. I don't recall why I rejected Ember, but MCM gave me problems. At the time, it didn't work well with TV episodes. Plus, it's major draw was automatic renaming, but when it pulled bad metadata, it gleefully renamed everything to the wrong movie name. In the end, I found it much, much simpler just to use the scrapers built into XBMC after manually naming my folders and files to match the contents of themoviedb.org and thetvdb.com.