We are currently working on stereo photography:
Using variational methods and rectification routines (and using Graff for processing images) it is possible to reconstruct disparity fields and also depth maps. Here for example we have a 3D portrait reconstruction:
It is comparable with the left frame of the stereo image:

In order to train a 3D face recognizer, we are looking for voluntieers, to shoot stereo portraits and reconstruct theirs face geometry. There are the first ones:
Till today we have found 36 volunteers. Our plans are to have one hundret.

After hard work, we glad to present 25 first reconstructed faces:
In total, till today 35 volunteers agreed to participate.

This morning at last we have built a 3D mesh, based on one of calculated disparity maps.
3D Face model with the correspondong texture:
And it's wire model: