3D Face Capture

Merck 3D Face

I am putting together a strategy for creating a bust of a donor who is supporting our tech build out. My idea is to do a 3D capture of their head and then fabricate it with the 3D printing hardware at Prince Labs.

Dr. Stephen Sullivan demoed the Canfield Vectra 3D capture system in his clinic for us. Scott Collins brought the surface into MeshLab for review.

The setup uses 5 camera arrays and produces a result in just a couple of minutes. There was pretty severe artifacting around my hair.

In order to create a water-tight printable object, Scott imported the incomplete surface into MeshLab and used the Poisson surface reconstruction filter on it.

Merck MR Brain

Ideal settings appear to be:
octtree depth = 9 (tile resolution)
Solver div = 7 (speed up processing in some cases)
Samples = 0.8 (higher = noisier, too low = overly smooth)
Surface Offset = 0.9 (1 or greater creates "hood" artifacts)

Dr. Sullivan suggests that a possible use case for topological filling may be computing facial symmetry measurements on noisy camera data. We need to import a clean surface back into Canfield to test.

Next step is to fuse it with my old brain MR, and see if my brain still fits into my skull.