PS- i found those exact rigs for 10 bucks with free shipping on ebay! NICE!
I have a background in Computer Science, so I might be able to give a different take on it. This is a computer vision paper that has been published in the journal IEEE Computer Graphics and Applications and from the reference in the video, it looks like it was presented at SIGGRAPH as well. You can get to the full paper as it is available publicly, but it doesn't look like they have made the software available. Given this is a research paper it is likely that the software they developed for this will still be very limited, and would be a long way off from a consumer product.
I think this would be mostly useful for ameratures wanting to make a fun video. I could see it as an addon to the popular vine app on iPhones for example. I don't think this would be very useful to anyone that is used to making stop motion animation. There would be a lot of limitations that would stop an animatior from using it.
Essentially what they are proposing is a software system that uses a very different workflow. You need to provide a realtime video of an object moving how you would like it to move. Your hand is basically a rig in this situation. You then need to provide an alternative source of information without your hand/rig in the same location. The software helps you do this by automatically suggesting keyframes that it wants you to provide and uses an onionskin video feed to help you recreate those frames. From those two sources of information it will automatically combine the two sources to provide a video without any hands/rigs.
The problem with this workflow is that it limits the animators freedom to animate. Basically you are only going to be able to move the object around. Additionally you need to be able to re-create the key frames that the software needs. This may also be a problem for a lot of animators as most people film forwards and would struggle to re-create a previous frame.
If they produced software that just did the automatic rig removal it could have some value for experienced animators. You would need to film a sequence with two exposures. Both exposures could contain rigs, but it would need to be moved so that there is no overlap between the rig in both pictures. This could be filmed forward in order like normal. The software would then be able to produce a single image sequence from those two with the rigs automatically removed. This wouldn't be revolutionary but it could be a timesaver.
thanks so much for the insight!