Cs192-26 final project

Aaron Chen Fall 2020

Project One: Light Field Camera

Depth refocusing;

We average shifted images by different alpha values in order to focus on different parts of the image. Shifts are done as described by the Stanford website- with the annotated distances. You can see in the following gif how it beautifully focuses from front to back.

alpha values were -20 to 3, but scaled by 0.9.

Aperture Adjustment;

By using fewer photos, selected by radiuses of 1 to 8 from the center of the given grid of images, we can emulate a smaller aperture. You can see the focus maintained in the back, and blurring in the front.

Summary: Awesome project! I love how we could create this cool effect from images that weren’t blurred at all.

Project Two: Image Quilting

Original brick texture:

Random selection of patches.


Quilting algorithm- overlap ssd penalty.

Texture transfer- results are poor, the running time was significantly slow so it was difficult to produce a more pretty output. Apologies. The way it works is by using the target image as additional info- we add to the penalty by the difference between target and the prospective texture.

Summary: I didn’t like this project as much, but it was pretty cool to see how effective simply using ssd as a penalty could produce cleaner image quilts.