CS194-26: Intro to Computer Vision and Computational Photography, Fall 2021

Project 3: Face Morphing

Angela Chen



Defining Correspondences


Image of me.
Image of Korean actress, Han Ye-seul.

Keypoints of my face.
Keypoints of Han Ye-seul's face.
Triangulation of the average of the points of my face and the points of Han Ye-seul's face.

Computing the "Mid-way Face"


Original image of me.
Original image of Han Ye-seul.
Midway face of me and Han Ye-seul.

The Morph Sequence

A 45 frame morph sequence from my face to Han Ye-seul's face.

The "Mean Face" of a Population

I used a set of 37 Danish faces for my population.

Some of the Danes morphed into the average Danish face shape of this set.


The average Danish face of this population.

My face warped into the average Danish face geometry.
The average Danish face warped into my face geometry.

Caricatures: Extrapolating from the Mean

To extrapolate my face from the average Danish face, I computed the target face shape as target_face_shape = (1 - alpha) * my_face_shape + alpha * avg_dane_face_shape. I then warped my_face_shape into the target_face_shape.

Here are the results with varying values of alpha:

alpha = -0.5.

alpha = 0 (just my face shape, no Dane).
alpha = 1 (my face shape warped to average Danish face shape).

alpha = 1.5.
alpha = 2.

Bells and Whistles

I attempted to make myself look more masculine.

Average face of Chinese men.

Shape morph.
Appearance morph.
Full morph.

All images besides my face were found on the Internet.