Part 1: Defining Correspondences

Using ginput in pyplot, I define pairs of corresponding points on the images of me and Ben Affleck. Based on the corresponding points, I compute the Delaunay Triangulation on each of the images.

Part 2: Computing the "Mid-Way" Face

To get the "Mid-Way" Face, I start with computing the average shape, i.e. the position of the corresponding points in Part 1. Then, I use the affine function to warp me and Ben Affleck to the average shape. At last, I average the colors, i.e. the pixel value, of the two warped images.

me

Affleck

"Mid-Way" Face

Part 3: The Morph Sequence

To get the morph sequence, I repeat the process in Part 2 with different weight for each image. One weight is for the shape(warp_frac), another weight is for the color(dissolve_frac). Both are essentially the parameter in linear interpolation.

Part 4: The "Mean-Face" of A Population

I use Danes/IMM face dataset to compute the mean face of men and women in the population (only the front faces with neutral expression). I use the similar method as in Part 2 with only difference of the weight of each image. Since there are not many women faces in the dataset, the mean face of women is not as smooth as of men.

men mean face

women mean face

I then morph some of individual faces into the shape of the men mean face. Above is original face, below is morphed face.

I then warp my face into the shape of men mean face, and warp the men mean face into my shape.

Here are some caricatures with extrapolating from the mean, which means alpha is no longer between 0 and 1.

alpha = -0.5

alpha = 1.5

Part 5: Bells & Whistles

Part 1: Smile!

I use the happy faces in Danes/IMM face dataset to compute the happy mean face of the population. By applying the shape difference, or the color difference, or both, I make my face also smile.

me

smile mean face

only shape applied

only color applied

both applied

Part 2: Morphing Music Video of Students in Class

I group up 22 students in class to create this morphing video. Here is the morphing of the previous student to me, and me to the next student.

Below is the complete video. Shout out to Zixian Zang for organizing the morphing video!

Link: https://youtu.be/O2p-U615Rzs