Making the invisible motion visible

Here is a new MIT video software, that constructs first a backbone of an object showing minor movements while then amplifying these movements to a ridiculous extent. Not clear what I try to explain? See a video at SPON. Although the idea behind is quite simple and even the implementation not too difficult I see a great potential for monitoring babies ( and even small animals ) as it is running in real-time. More videos at, yea, yea.