I used python's cpselect to define my correspondences.
After creating the Delaunay triangles, I did an affine transformation from the desired midway face to both my face and george's face, interpolating the result to get the corresponding colors on the midway face. These gave me the third and fourth images. After averaging these images, I got the final image.
I created a morph function and created a morph sequence, slowly transitioning from the 0th to the 45th frame. I set my warp_frac equal to my dissolve_frac.
For this part, I averaged the values of all smiling men in the Danes dataset. It was the same process as the above, making a transformation from the average shape to each of the faces and interpolating, then finding the average of all the results. I then morphed my shape to the average, and the average shape to me.
Instead of doing a transformation from the triangles in the average face to mine, I multiplied my triangles by 1.5 and subtracted 0.5 times the average triangle, to exaggerate my own features. The following was the result.
I tried to change my gender by morphing my face into the supposed average female, found from google images. The third image is only changing the shape, the fourth image only changes the appearance, and the last image morphs the shape and the appearance of both in equal proportions.