Colour

Explanation

When you run the demonstration, the photograph in the illustration is subjected to progressive pixel averaging. The image is divided into progressively larger squares. Each square is replaced by a uniform block of colour. The new colour is calculated as the average colour in the corresponding square in the original image. The average colour is found by taking the average value in each of the red, green and blue colour channels.

This process of averaging out the colours is used by image processing programs when they carry out automatic colour correction. The image is adjusted so that it averages to a mid grey, at which point the colours are said to be balanced.

Click again in the illustration when the demonstration is complete. The new image was produced in Photoshop 7 using Image Adjustments Auto Color. To my eye, this produces a slightly harsher and colder image that does not convey the original scene as well.

Click again in the illustration to run the demonstration on the adjusted photograph. The adjusted image does indeed average to mid grey, not the grey-brown of the unadjusted version.

Observations

• Automatic printing devices assume that any picture will average out to mid grey like the artificial combinations obtained by apply the formal rules of harmony. This assumption is close enough to the truth to have allowed the old one hour photo labs to churn out prints using automatic colour correction. It also explains why you could expect a white cat to look slightly magenta if you photographed the cat on a green lawn and had prints made at a one hour lab.
• You can see the effects of progressive pixel averaging on your own photographs. Download the source code for the demonstration. The file contains instructions on how to use it.