In my thesis I developed a novel method for Caricature Generation. This method was implemented in a program which takes images of faces as input and automatically produces caricatures as output.
The method relied on the notion of “anti-face” in conjunction with the use of unbiased distortion functions. These two concepts constituted the basis of a mathematical set of rules that can be used for caricature generation purposes.
Simply put, the anti-face is the exact opposite of a given face, as depicted below:
x: A given face
μ: Average face
x’: Anti-face of x
These are some sample results of a program that implemented the method:
Caricatures generated by the program
By deploying a reverse variant of the method one can drive an input face towards the average face. Studies show that the average face is commonly considered to be attractive. Thus, in a sense, the method can also be utilized in order to attain beautification.
Example of the same program adjusted for beautification:
Left: Initial face
Right: Beautified face