Beans |
Lego Truck |
|
|
|
|
|
|
|
|
Chengzong Ou <aai>
In this project, I used two sets of images from the Stanford Light Field Archive to perform both Depth Refocusing and Aperture Adjustment.
The refocus the depth of a Light Field image, we first need to obtain their offset (u,v) values. This is done by parsing the filenames and reading the embedded coordinates. Since there are 17 cameras in both horizontal and vertical directions, I chose (8, 8) to be the center camera. I then calculated the offest of each camera from (8, 8). Multiplying these offsets with an alpha value between 0 and 1 and shifting the images using these offests recreates refocused images.
The following are two sets of images with refocused points.
Beans |
Lego Truck |
|
|
|
|
|
|
|
|
|
|
|
|
Aperture Adjustment is done by only including a subset of images. In this project, I defined the aperture number as including images that are at most aperture away. For example, aperture = 1 means that a 3x3 configuration with vertices at cameras (7,7) and (9,9) are included.
Beans |
Lego Truck |
|
|
|
|
|
|
|
|
|
|