When it comes to post-processing, there are options to remove noise as well as to add noise. This is yet another contradiction that every photographer faces once in a while.
Noise in photography refers to the small specks which become visible at high ISOs. Sometimes noise can also refer to areas of discoloration. In the film days, the higher ISO films had ‘grain’. Noise is unwanted but grains can add character to the photograph. This statement should clear all the doubts. Read it again if you want to. Unfortunately a lot of image editing programs call this ‘grain’ also noise and thus the confusion.
The sensor is an electronic device which converts light into electrical signals. This is recorded as a photograph. When the light is low, the sensor along with the supporting electronics has to work hard to get the light converted to electrical signal. This causes artifacts too. This is noise. It can be due to various factors at play but at the end it robs the image of the clarity and even causes change in saturation or color balance. With advancement of technology, cameras are now capable of capturing reasonably fine images at insanely high ISOs. Such ISOs were unthinkable with the film.
Some dos and do nots for noise reduction-
Use a low ISO whenever possible… this is easy and yet people forget or ignore this. Even a single stop shift in ISO to the lower side can be helpful. A word of caution – keep a check on the histograms to avoid underexposure. Switch off the camera in between shots and keep it off until the next planned shot. This gives time for the electronics to cool down. Did you know that the camera is less prone for noise in winters than summers? Switch off the long exposure noise reduction. I have never seen this feature work.
The specks are not equally distributed in the three colors. In most sensors, the blue and green specks are more than the red ones. So instead of reducing all color specks, try to reduce noise for only these colors if your post-processing software has this option. The size for noise reduction should be kept small enough just to cover the specks. Remember reduction in noise also means loss of some details. The aim therefore is to reduce the noise while keeping the details intact. The aim is to get an acceptable image at the end. Do not loose sight of the forest while looking for the trees. When you are removing noise at 100% or higher magnification, keep coming back to view the whole image for a better understanding of the overall photograph.
Next comes the noise that is in the same color as the subject but it varies in intensity. This is called luminance noise and some programs have dedicated option to reduce this kind of noise. This should be done after removing the colored noise specks.
As I had said before, removal of noise causes some loss of details too. Sharpening can be applied to cover some of the loss. Make sure not to overdo the sharpening.
Addition of artifacts to replicate film grain is called ‘addition of noise’. This is what makes the whole thing confusing. For clarity sake, this should have been called ‘grain’ everywhere but then not all programmers are photographers.
Addition of grain makes the image look sharper. Grains also add a character to the photograph. This may have been a left-over in our subconscious since the film days but it does make the photographs look more pleasing. Another word of cautions – if you are planning to submit your pictures to stock-image sites, keep the addition of grains low. Their algorithms can sometimes reject a photograph on technical grounds disregarding the overall artistic impact the grain would have had.
For some more pointers on post-processing – Post Processing RAW. This link also contains a photograph that I had shot at 4000 ISO after noise reduction.
(A confession – I rarely click photographs at very high ISOs. I do not enjoy sitting long hours, post-processing. Inserted the photograph here to add color to the article.)