Brian Zhu (brian_zhu@berkeley.edu)

Defining Correspondences

Original images for morphing:

With keypoints added:

Taking the average between the corresponding keypoints and computing a Delaunay triangulation:

Computing the "Mid-way Face"

Original Images:

Mid-way face:

The Morph Sequence

The "Mean Face" of a Population

I used the Dane dataset to calculated an average face.

Keypoints used in the Dane set:

I also added keypoints at the corners to prevent the triangulation from cutting off too much of the image. Triangulation of average keypoints across the entire dataset:

Average face:

Morphing the shape of dataset faces to average face:

Keypoints used for morphing between my face and the average face:

My face warped to shape of average face:

Average face warped to shape of my face:

Caricatures: Extrapolating from the Mean

Warping my face to 1.5 * my_keypoints - 0.5 * average_keypoints:

Warping my face to 2.0 * my_keypoints - 1.0 * average_keypoints