CS 194-26: Computational Photography, Fall 2017

Project 5: Lightfield Camera

Tom Cheng, cs194-26-aev



Overview

Simply defined, lightfields are a set of images of an object. These images have the same angle. However, they are positioned slighly "off-centered" from the center shot. Using this lightfield data, we can actually write code that manipulates the photo even after we've taken it. In this project, we explore lightfield depth refocusing and aperture adjustment.

Depth Refocusing

Depth refocusing involves switching our focus from one part of the image to another. To do this, I determined the "center" photo. Afterwards, I calculated the offsets to shift the image by. Lastly, I took the average of all the shifted images. I would do this for every value of c (-0.8 - 0.8, with a step size of 0.05), which corresponds to a different "depth". Below, I've displayed my results. As you can see, the focus moves from the top right corner to the bottom left corner in a diagonal.

Refocus

Aperture Adjustment

Quite interestingly, we can actually adjust the aperture that was used to take the photo. We can perform this operation by selecting specific images and then averaging those selected images. So, how do we pick which images? Imagine a grid of images with the center at (0, 0). An aperture adjustment of radius 1 would be selecting all the images surrounding that center. In particular, we'll be averaging a "3x3" square of 9 images.

Aperture (Radius: 0 - 8)

The End.

Thanks so much for reading!