The function is run according to the readme given. The basic morph took an input of two sets of control points, computed a triangulation using the delaunay matlab command on the median, and interpolated between the two. An affine transformation was computed for each of the corisponding triangles, inverted, and used to interp2 with cubic interpolation to find the correct morphed image.
The morph came out very well, since the hairstyle though the backround changes color. The pictures however where taken from two perspectives with regard the body position. this leads to disconuity in the neck. that the neck is not marked for wwarping means it can't be shifted correctly. The hair is quite interesting moving from combed to puffy.
Combined the images morphed to the average geometry is very good looking :) A good representation of the class, though not such a good representation of the world at large. The smiling mouth on my picture caused a odd effect of exagerating it's size on respective morphes. Average geometry makes it small, while using average fave on my geometry blows up the mouth.
The difference between the mean and my own geometry was used to created an exagerated geometry. It looks pretty goofy. In addition an odd bug created the second image which appears to be an even better exageration.
