HomeCulture and EntertainmentServices (only Italian)

Milanofotografo - Noise reduction through overexposition and correction during development of the raw file

Since digital photography permits to correct the exposition during postprocessing, as if it was possible to go back in time at the moment the picture was shot, I decided to investigate the possibility of reducing noise by overexposing. This because the photon counting follows the Poisson statistics, according to which the signal to noise ratio, apart from forward factors like amplification, tension oscillation, etc..., is proportional to the inverse of the square root of the number of the incoming photons.
Therefore the more photons it is possible to collect, the better should be the signal to noise ratio, indipendently from the iso settings. To test this theory can be applied also to the real case we took three pictures at 400 ISO (same aperture): one with the correct exposure time, one overexposed of one stop and one overexposed of two stops. Furthermore, to get a further comparison, the same subject was shot at iso 100 with the correct exposure. In the following the risults. First the 400 ISO picture with correct exposure, second the 400 ISO picture with one stop overexposure, third the 400 ISO picture with two stops overexposure and, last, the 100 ISO picture with correct exposure. Clicking on the pictures it is possible to download the complete photo. All the pictures were taken with the Fuji s100fs. Unlike in the photgallery no noisereduction was applied.


ISO: 400, Diaframma: 4.44, Tempo: 1/500
Scarica immagine completa


ISO: 400, Diaframma: 4.44, Tempo: 1/250
Scarica immagine completa


ISO: 400, Diaframma: 4.44, Tempo: 1/125
Scarica immagine completa


ISO: 100, Diaframma: 4.44, Tempo: 1/125
Scarica immagine completa

Clearly the last two pictures show the same amount of noise. This indicates that our theory is correct. The differences in contrast are due to the different dynamic range, which decreases with the increasing of the ISO.
Which one is therefore the right strategy when taking pictures, based on the described results? Unlike what happened in the analogical photography one should try to collect the highest number of photos, limited by the need not to saturate any part of the sensor. Furthermore one should use the lowest ISO setting, in order to maximize the dynamic range. The correct exposure can then be obtained by correcting during postprocessing.
What can be the contraindications and the risks?
  • One risk is to burn the brightest parts of the picture. Ones burnt these parts are lost.
  • It can be difficult later to choose the correct exposure in order to recreate the scene shot with the correct illumination. It could be even useful to take two pictures: one in the classical way, as a reference, and one in the abova described mode to get the best image quality.