Today’s “Ask Tim Grey” email (which you can sign up to receive here: http://timgrey.me/greylists) resulted in a great deal of feedback from readers, so I thought I would publish the full question and answer here on the GreyLearning blog. The question related to the notion of the camera’s ISO setting not really being the cause of noise in digital photos. That notion is mostly true, but requires clarification. Here is my effort at providing that clarification:
Today’s Question: I just saw an article that said raising the ISO setting does not actually increase noise in a photo, but instead shorter exposure durations cause noise. This doesn’t match what I’ve always read. What are your thoughts?
Tim’s Quick Answer: It is mostly true that raising the ISO setting isn’t the true cause of noise in a digital camera. However, it is important to keep in mind that the article in question specifically related to astrophotography. For more terrestrial forms of photography, it is still generally safe to assume that a lower ISO setting will translate to reduced noise levels.
More Detail: As I’ve said many times, noise is the opposite of information, and in the context of photography light is the information we’re dealing with. Thus, less light will translate into more noise. This is the foundation of the “expose to the right” principle, which calls for capturing photos that are as bright as possible without losing highlight detail in order to maximize detail and minimize noise.
However, this does not mean that you should use a high ISO setting to minimize noise. Quite the contrary for most photographic scenarios.
Raising the ISO setting will require that you either use a faster shutter speed or a smaller lens aperture opening in order to reduce the amount of light reaching the sensor. The amplification triggered by the increased ISO setting will then compensate for the exposure.
However, raising the ISO setting really translates into (potentially dramatically) underexposing the image, and then amplifying the capture information in the camera. The underexposure is indeed the key cause of noise, but that underexposure was caused by a higher ISO setting. So the two are related.
Furthermore, this issue is more nuanced than the article suggested, because there is a big difference between underexposing with versus without an increase in ISO. If you use the exact same shutter speed and aperture settings at a low versus high ISO setting, you will see more noise (and less detail) in the capture with the low ISO setting. This is an indication that the camera is able to do a better job of brightening the image (through amplification of the signal) than our computers are able to do by simply brightening pixel values.
But again, more light will help ensure the lowest noise levels. Thus, you generally want as much light to reach the sensor in the camera as possible. That, in turn, means keeping the ISO setting at the minimum setting, so that you will use a larger lens aperture and/or a longer exposure duration to compensate. That results in more light, and therefore less noise.
So for most photographic scenarios, it still holds true with most cameras that you want to use the lowest ISO setting.
The reason a different approach to ISO makes sense with astrophotography is that you generally don’t have any flexibility when it comes to shutter speed and lens aperture. You may be shooting with the lens aperture wide open, and the shutter speed at the longest exposure duration possible without introducing star trails. If you need more signal, your only option is to increase the ISO setting. As noted above, a higher ISO setting is generally preferable to a low setting when all other factors (shutter speed and lens aperture) are fixed.