Computational Photography

HW 5
Daniel Jacobs

The project was to create a pinhole camera. The camera, shown below, was constructed from a shoebox. One inside wall was papered with white paper, all others were papered with black paper. Opposite the image wall two holes were created, one for light to enter the box, and one for a digital camera to take a picture through. A secondary box was built around the camera hole to reduce light leakage. The imaging hole can be covered with a notecard to reduce the aperature size.

Outside

 Inside
The smaller the hole, the clearer the image produced by the camera, and the larger the hole, the brighter the image. Shown below are image with hole diameter about 1mm, 4mm, and 10mm. The 1mm hole does not let enough light in to form an image, while the 10mm hole blurs the image significantly. Unfortunately, the camera I was using had poor low light performance, and was only able to produce images when the subject was a light.

1mm

4mm

10mm

1mm

4mm

10mm

4mm

4mm

4mm

4mm

Here is an attempt at light painting. The goal was to draw a trident using a flashlight and the camera set to long exposure. The shape is clearly visible, but it appears as a bunch of blurred dots, rather than a continuous line. This reveals part of the issue with the camera. In addition to poor low light performance, the long exposure mode on the camera is actually taking many discrete images, and then summing them together. I suspect the camera is simply taking a video at 15-30 fps. This means that any light below the threshold for a 1 in the image over a small fraction of a second never registers.