What technique would be the best for 3d shape acquisition considering the following aspects: Let's say that that the shape that you want to scan is into a specific category, like a car, or a bike etc.
You have a fabric material, like a blanket or something that you can cover the object with before taking the pictures.
Given the fact that you can choose exactly the pattern on that blanket, and use the way it modifies or distorts when it is covering the object, what approach would be the best to get as much info as possible on that shape from a small number of pictures.
I was thinking that the pattern on the blanket is kind of similar in concept with a pattern that a projector would create. But it is not really like that because the parts of the object which are further from the camera would not increase the size of the pattern from the blanket, while the projector would cause an increase there.
If you were to add some specific shapes that were easily and uniquely identifiable, maybe multiple view geometry would work, but I don't really know too much about it and I kind of doubt its precision.
Another approach that I was considering was to make the pattern on the blanket consist of only vertical and horizontal lines, and "interpolate" the resulting angles in order to construct the surface. A similar example would be in the following picture:
Keeping in mind that my knowledge on the subject is very limited - I just read about this a while ago- what approach would be recommended? Why?