Andy Wu, Fall 2020
This project explored multiple image processing and methods through the manipulation of the frequencies in images, through the use of various filters and stacks.
The original images are
Then, we combine the two faces to get the mid-way face.
We can linearly change the morph coefficient to achieve the morph sequence.
I used the Brazil FEI Face database for the face images and labels.
Shown below are some examples of the before and after of faces morphed to the population average.
Shown below is the average face of the dataset.
Using the average face, we can morph the average to match the shape of my face,
or morph the shape of my face to match the average.
Finally, we can extrapolate from the mean to create a caricature.
Using the two different averages of not smiling and smiling faces, we can attempt to change my face from not smiling to smiling.
I think the transformation turned out poorly since the labeling of points on my face was not exactly matching the average change from not smiling to smiling, resulting in a visual artifacts that make the face look bad.