I couldn't watch it yet but I assume it's only a sideview of the frames? What you actually need to do homebrew pseudo motion capturing is a sideview and a front view of the same scene (or a highly similar one). This way you'll align the bones first to the sideview frames and second to the front view, using the frames as background images of your animation tool. This should result in top quality quasi-motion-captured animation.
It may look silly, but it's easy to use a lowcost camera and record this by your own. </offtopic>
It may look silly, but it's easy to use a lowcost camera and record this by your own. </offtopic>