Face Morphing

I asked my little brother for a selfie so I could turn him into George Clooney. He asked for Stephen Curry instead. Below are the original images of his face and Curry's face that I used for the morph.

Here are the average shape points and Delaunay triangulation of those points mapped onto my brother's face and Curry's face.

Here is their average face with morph_frac set to 0.5 and dissolve_frac set to 0.5.

Here is the GIF morph from my brother's face to Curry's face, both slow and fast. The gif contains 50 images in which I increased the warp_frac and dissolve_fract by 0.02 from 0 to 1.

Average populations

I used Michael M. Nordstrøm's free data set of faces to contruct an average. Since the collection featured 40 faces at 6 different angles, I chose 40 images from the original 240 with consistent poses to do the face averaging (I chose the first pose).

I took the average of the 40 face shape points provided, and morphed each face to that average shape. Here are some examples of faces before and after they were morphed to the average shape.

After calculating the average face shape, I morphed every face to that shape and averaged them together. Below is the result of averaging all the faces.

I then tried morphing my face into the average face shape. Here is the before and after.

Here is the average face shape warped into my face shape.

To caricaturize myself, I extrapolated my face first more towards my own shape with alpha = 0.3, then extrapolated my face more like the average population with alpha = -0.7. Below is my face more like me, then my normal face, then my face more like the population average.