CS 194:-26 Image Manipulation and Computational Photography, Fall 2018

Project 5: Light Fields

Jun Zhou, CS194-26-abm



Depth Refocusing

The lightfield data consists of images arranged in a 2D grid where each photograph is taken from the position in the respective grid position. By shifting an image from their position to (8, 8) to compensate for the parallax effect, we can line up our two images at a specific point and blur out the rest of the background. We extend this to all our images and stack accordingly for the 17 x 17 photographs, then scale by c (from 0.0 - 1.0) to achieve the depth focusing effect.

Aperture Adjustment

Taking the same principle of compensating for the parallax effect, we instead take a subset of the images to average over. By taking a 2x2 6x6 10x10 and 14x14 centered around (8, 8) we get the aperture effect on our camera.

Summary

It was amazing to see such a simple concept blossom into fluid aesthetic effects. Typical cameras use many hardware features to enhance their effects, and using a hardware technique integrated with software allowed me to see the beauty of computational photography.