Project 5 Lightfield Camera
Amy Shan
Overview
In this project, we implement depth refocusing and aperture adjustment using datasets from the Stanford Light Field Archive. The idea is that if you take multiple pictures over a regularly spaced grid you can achieve complex visual effects using simple shift and averaging operations.
Depth Refocusing
Imagining the 289 images provided in the Stanford dataset as a 16x16 grid, we choose the center (8, 8) as the image to normalize the rest of the images in the grid to. We choose a constant C as a scaling factor and use it to scale the difference between the center (8, 8) and each of the images (u, v): C*(8 - v), C*(8 - u). I chose C from a range of 0 -> 5.
Aperature Adjustment
To simulate the effect of an increase in aperature, and therefore the blur, we increase the radius within which we average images in thereby increasing the number of images. I chose the same center image and chose a range of 0 -> 6 for the radius.
Summary
I learned that light fields can be created computationally if a large number of images are first obtained.