Fig 1.
Each photographic exposure generated a stack of up to 12 images, each in focus at a different depth plane.
Fig 2.
Steps to determine the in-focus area for the first (closest to the camera) image in the stack.
a) The original 20x20 pixel depth map was (b) up-sampled to 1080 pixels and smoothed with a Gaussian, so that (c) the distance in depth of each pixel from the first image’s focal depth could be computed. d) The local standard deviation of the depth image, computed over a 165x165 pixel region, was used to identify areas of uniform depth. e) The product of c and d: the white circle is centered on the location where this product has the lowest value (closest depth and lowest variance), to indicate the position of the in-focus patch at this depth.
Fig 3.
Example of extracted in-focus patches for each focal plane.
Left: in-focus patch for the first image in the stack. Right: locations of in-focus patches on the depth map.
Fig 4.
Blur in a conventional lens system.
An object at distance Z from a lens with focal length f and aperture A is imaged on a sensor plane at distance s. The object will be blurred over a circular region with diameter C.
Fig 5.
Dioptric depth separation between image planes for the ten light field pictures in the study.
Black circles represent the mean dioptric distance for each image in the stack. Error bars are ±1 standard deviation. Dotted lines represent the linear fit (coarse dots) with 95% confidence intervals (fine dots).
Fig 6.
Example of one trial from Experiment 1.
Each trial began with a blue fixation cross presented in the center of a grey screen for 1 s. Two stimuli were presented in a circular window with abrupt onset and offset for 200 ms, separated by a 500 ms interval containing only the fixation cross. The subject was required to fixate the cross and to indicate whether the first or second patch was farther, using any available cues.
Fig 7.
Example of one trial from Experiment 2.
While fixating on the central cross, a subject was required to indicate which patch, the left or the right, was farther, using any available cues.
Fig 8.
Discrimination sensitivity d’ for each condition in the (a) temporal 2AFC task and (b) spatial 2AFC task, averaged over the four observers in each experiment. The graphs show group bootstrapped distributions of discrimination d’ for the four experimental conditions. Vertical bars represent the true group means.