Second: I worked alongside a brilliant team with rare knowledge of geometry reconstruction. We took the 70 images of people, algorithmically removed the backgrounds and then used a proprietary process to create a mesh geometry from the intersection masks of those shapes in 3D space, which all had to be carefully calibrated on the regular. That geometry was smoothed to create a valid smooth human mesh. We then created a texture which would be mapped to the physical geometry of the mesh. All lighting and camera sensors had to be carefully measured and corrected, with ample algorithmic color smoothing along the seams of this texture. Then we ran our proprietary auto-rigging algorithm on the mesh, and exported both a 3JS compatible obj+jpg as well as a Unity bundle. The entire process was optimized to take under 60 seconds. This meant that you could step out of the booth, watch the magic happen, and then immediately put on a Vive and interact with yourself in a mirror using your fully articulated new avatar.
Third: VCs were unwilling to fund something with a hardware component
Fourth: sigh
