Lightfield Cameras

Nerissa Lin CS194-26-adv

Overview

Having seen the Lytro camera a few years ago, I was captivated by the technology and the ability to apply edits to these photos after the moment of shooting the photo, something that was unheard of. Exploring this tech in class, we learn that with enough data about the scene, coming in the form of photos taken at multiple different points, we are able to derive new information from the photos and manipulate them to create scenes that are focussed in different points. Beyond that, we can imitate different aperature sizes given this same data set and we show these results below

Depth Refocussing




Depending on the alignment of the photos in the Stanford Light Field Archive set that we choose to work with, we can mimic different focal points. Without shifting the image we seem to arrive at an image that is focussed near the top but the outsides are out of focus. Similarly, shifting the image by different amounts show focal points at other places in the image like perhaps near the bottom when we shift at 4 times the displacement. At a "perfect" alignment of shift in the supposed displaced amount, we get the center focussed. I found that here, the best scale ranged from -1 to 4 where in this range, the shift of focus is the most notable and easy to observe. Below we see the example of the image on the left from just an average of all the images, and the one on the right, with a clear center focal - being the one that is aligned.

Aperature Adjustment

We have information about the displacement of the camera at each of the points of shooting, so interestingly enough we can try to mimic the concept of aperature size using this data. Starting out with a single image, we can choose the one centered at 8,8 - we get something that would mimic a small aperature - resulting in only straight rays, mimimal blurring. However we can expand our aperature size to be something of a larger radius and to mimic this with our data, we grab photos of the image displaced radially around the center image within this radius amount. Doing so, we average these photos and get the angled ray data from the displaced images, as if we had increased our aperature. We see this result largely in the fuzziness introduced at the edges, and slightly stronger focal point near the center.

Aperature of size 0 and 1

Aperature of size 5 and 8


Summary

Seeing the results of this project was really cool because it helped me better understand the various functions of the camera - seeing how focal point can be shifted across an image was enlightening from a technical perspective but was also cool to help me identify new ways to look at some of the images that I already had preconceived artistic ideas about. Better understanding aperature was cool too since it let me appreciate the artistic value of it more.