As part of CS 194-26 (Computational Photography) at UC Berkeley, I created a collection of computer-manipulated images using techniques such as Poisson blending, image warping through the affine transformation of a Delaunay triangulation, Sobel filtering, and more.
You can see a sample of what I made in the gallery below, but you can also see some interesting results here and here .
A "caricature" of myself, created by finding a transformation between an "average" face and my own, and then extrapolating that to exaggerate my features more.
A "hybrid" image which looks mostly like a trash can when viewed up close, and mostly like a human face when viewed from far away. It was created by applying a high-pass filter to the trash can and a low-pass filter to the face, and combining the two images.
An image created using a technique called gradient mixing - essentially, the image gradients are combined in a specific way to ensure the transition between the two is not too "rough".
Another example of gradient fusion, this time using a technique called Poisson blending. Mixed blending works better when you want to preserve the background image more, but here Poisson blending works just fine.
An image with its depth of focus artificially modified. It was created from a series of pictures taken from a camera array known as a Lightfield camera.
An example of how we can artificially modify the apparent aperture of a photo taken by a Lightfield camera. The aperture determines the amount of light let in when taking the picture, affecting the focus.
