Project 3: Face Morphing

By Yiyu Chen

▸ Part 1: Defining Correspondences

Correspondences

▾ Part 2: Computing the "Mid-way Face"

This is what each picture looks like when shaped into the Mid-way Face.

Image a to Mid-way — Image a (No smile) to Mid-way

Image b to Mid-way — Image b (Smile) to Mid-way

Mid-way Face — Combined result: Mid-way Face

▸ Part 3: The Morph Sequence

To make the morph more natural, I used different functions for shape distortion and cross dissolution. For the shapes I used a linear function, but for the cross dissolve I used a quadratic easy in out function with an “S” shape. Here are what the functions like:

Function curve

result:

Animated GIF

▾ Part 4: The "Mean Face" of a Population

Here is the "Mean Face" of the population:

Mean Face

Mean Face 2 — Mean Face with Smile

Mean Face 3

Here are some faces morphed into the mean face:

13a_ori — Original face 13

13a_morph — Morph to Mean: face 13

53a_ori — Original face 53

53a_morph — Morph to Mean: face 53

103a_ori — Original face 103

103a_morph — Morph to Mean: face 103

106a_ori — Original face 106

106a_morph — Morph to Mean: face 106

153a_ori — Original face 153

153a_morph — Morph to Mean: face 153

183a_ori — Original face 183

183a_morph — Morph to Mean: face 183

ori face — Original face

shaped face — my face warped into the average and the average face warped into mine

▸ Part 5. Caricatures: Extrapolating from the mean

Here's a caricature of my face, this is what I get when I set warp_frac to -0.8

Caricature

▾ Part 6. Bells and Whistles

▸ Age Progression

I picked a photo of myself from different ages and make a movie that shows how it changed over time.

▸ Gender Changes

I'm using the mean face I created earlier here. I convert the mean female face to male and the mean male face to female.

Change gender