With digital capture the golden rule, as always, is not to blow the highlights. Let’s think about what is happening when we make a digital capture. When light hits the sensor, the number of photons (light particles) hitting each pit on the CCD is counted. The more photons that hit the pit the, brighter the resulting pixel in the image. But this is not the end of the story because there is an inherent randomness involved. Even for areas that ‘should’ be exactly the same colour the number of photons counted will not be exactly the same. This leads to noise. The statistics of the situation show that there is a higher percentage variation in the number of photons counted in dark areas than light areas.

Conclusion: there is more noise in the darker parts of the image.

As we all know, to increase the exposure by one stop we need twice the amount of light (twice the number of photons). To increase the exposure by two stops we need four times the amount of light. This has not changed since Fox-Talbot’s day. Unfortunately the sensor in the camera is a linear device and our cameras have 12 bits available to record all different possible tonal values. Putting these two facts together means that the brightest stop needs to record twice as many values as the next brightest, which in turn needs twice as many as the next, etc.

Conclusion: the brightest stop in the image contains fully half of the information in the whole picture.

Putting these two ideas together:
1. The brighter the area in the picture, the less noisy it will be.
2. Most of the information in the picture is at the brighter end.

This means that if we want to reduce noise to a minimum and keep as much data as possible for post processing we need to get as much of the image as we can in to the brightest few stops.

Copyright © Greg Wright 2006 - All rights reserved