3D for the masses?

Came across this new interesting startup Fotowoosh. Basically these are guys from CMU that I have come across before (look here ) who use something they call "geometrical contexting" to create 3D images from (supposedly) any single image. Indeed it's quite cool - but I wonder what the real value of such technology is since, from consumer level perspective it only works well on well defined shapes and lacks the fine structure. More notably it seems to distort humans - which probabaly make the most part of private familial photo collections (notice that in their demo there is one shot of a street with a man on the sidewalk...) so I am unclear as to their lasting merit.
Here is the TechCrunch link.

MC