This project involved "morphing" faces into each other. The underlying technique is affine transformations. By defining points on both faces, with each point on one's face having an equivalent point on the others face, we can find weighted averages of these points. Finding the Delaunay triangulation of the average points and using affine transformations to transform points from the source faces to the points in a "mixed" face for each triangle, we can develop a weighted average of the two faces. We can do this for many weights, until we get a morphing sequence from one face to another.

I morphed my face into the famous basketball player James Harden's face. I included points that defined the shape of our beards.

I took the Dane dataset and morphed all the faces to the average face points, and the following was the result:

I then tried to project my face onto the mean face, and vice versa:

I took the difference between the mean face points and my face points, and subtracted the result twice from the average face to produce a caricature face data points. I put my face on it, and here are the results. Sadly, I'm pretty sure I can make that face without this technology: