Overview
In this project, we used a lightfield of an object, which were images taken of an object over a consistently spaced grid to produce the lightfield, as demonstrated below.
Depth Refocusing
In depth refocusing, we use a light field which gives us images of an object from slightly different camera positions. To refocus, we calculate the distance from the center of each image's u, v coordinate. Then, we shift the image by c * (distance u) and c * (distance v) for some c between [-1, 1]. Finally, we take the average of these shifted images to get a refocused image. Here are our results at varying values of c.
c = 0.1
c = 0.3
c = 0.6
gif with c in [-0.1, 0.7]
Aperture Adjustment
For aperature adjustment, we follow a similar logic that we used in depth refocusing. However, to adjust the aperature, we take a selection of images that are within some distance from the center image. Here are our results at varying radii. For example, radius = 4 means that all images we have selected are at most 4 away from the center. Here are our results at varying radius levels.
radius = 1
radius = 4
radius = 7
radius = 10
gif with radius in [1, 15]
Summary
I found this project to be really interesting! It was cool being able to essentially simulate focusing on an object or adjusting aperature size with just a set of images that were taken at various camera positions.