the latter is definitely the long term vision. Today we do capture the elements, but do some sort of 'automated' (tree traversal) to produce the video. Would definitely love to give future users even more fine grained control