Project 4: Face Morphing

CS 194-26: Computational Photography

Caroline Moore Ė cs194-26-aew


For this project, I manipulated images of faces to find warpings between individual faces and between those faces and an average face of a population.

Part 1: Two Faces

For the first part of this project I found the average of my face and by (fraternal) twin sisterís face.

My face is on the left, my twin sisterís face is on the right, and the average of our faces is in the middle. This worked pretty well, except her ear was showing and mine was not so the ear is covered by translucent hair in the average image. The backgrounds of the images are also very different, so it just ends up looking washed out.

I also made a morph sequence turning her face into my face.

Part 2: Mean Face

Next, I found the mean face of a population. Specifically, I found the mean face of several Danes. To do this, I first found the average set of points for all the faces. Then, I warped each face in the set to those of points. Finally, I averaged these warped images.

I also found the mean faces of a few subsets of the population. To start with, I found the average faces of the Danish men and the Danish women.

There were significantly more images of men than images of women in the data set, so the average male face looks better than the average woman face.

I also found the average of the same Danes but with happy expressions.

Using the average happy face and the average neutral face, I turned my neutral face (left) into a happy face (right).

Using the average neutral face, I also warped my face to the shape of the average face (left), and warped the average face to the shape of my face (right).

Then, I used the average neutral face to create a caricature of my face (right).