Yiwen Chen

## 1.1 Defining Correspondences¶

In this section, I defined 43 pairs of correspondences manually using ginput. Then I computed Delaunay triangulation with the average of each pair.

Original images:

Annotated images:

## 1.2 Computing the "Mid-way Face"¶

The following mid-way face is computed by

1. Computing the average geometry by taking average of each corresponding pairs
2. Computing affine transformation for each triangle from one of the original image to the mid way image. Use the inverse transformation to colorize points in the mid-way image.

## 1.3 The Morph Sequence¶

Similar to how the mid way image is constructed, for each a = 0.1, 0.2, 0.3, .. 0.9 a frame is constructed using a as both warp fraction (shape) and dissolve fraction (color).

## 2.1 The "Mean face" of a population¶

I used 6 ”happy” expression female faces from the Danes dataset. The original images are as follows:

To compute the mean face,

1. Compute the average face shape by taking average of all corresponging points given to us
2. Morph each of the faces in the dataset into the average shape.
3. Take average of all morphed face

Here is an example of step2

original:

morphed image:

The average face of ”happy” expression female faces:

## 2.1 From portrait to average and the other way around¶

Here is my original portrait:

The result of my face warped into the average geometry:

The result of the average face warped into your geometry:

## 2.2 Caricatures: Extrapolating from the mean¶

A caricature of my face is calculated by weighing my face with 1.5 and weighing the averge face with -0.5 for both shape and color:

In [ ]: