I computed the mid-way face between Donald Trump and Kevin (from The Office). To do this, I picked roughly 50 corresponding points on each face, triangluated the points, performed an affine transformation and output half of each transformed image.
I did the same as above, but instead of `0.5` as the warp and dissolve fraction, I used the corresponding frame (factor of `1/45`) and wrote each image as a frame of a video.
I got the average Danish Face by using the points given in the data-set to morph and display each face together for an "average" of all the face in the data set
I warped my face to the average Danish face and vice verse. This uses the same morphing process as above, and I played around with the warp/dissolve fractions to find a better output. The warps are not as smooth as I would have liked because I did not put enough corresponding points on each image.
My Face |
Average Danish Face |
My Face Warped To Average Face |
Average Face Warped To My Face |
I made a caricature of my face by warping my face with the average danish face and extrapolating the vectors between corresponding points. I play with the amount of extrapolation as seen in the images below.
Caricature 1 |
Caricature 2 |
I warped and merged my face with the average NBA player's face (found on the internet). Again, I played around with the warp/dissolve fractions to achieve a nicer result.
Original Face |
Average NBA Player's Face |