I would do it in a way you have explained it. Mask the video and arange it in a way you like. Anyway if you are looking to stylise your video see my tutorial on embryo.me.uk about crossprocessing. Use only chunk of video you need for your animation. As you have said motion does not like to handle a massive video files. Maybe use Prores codec, not HQ just Prores. File will be small while retaining the quality.