The role of ISO in digital photography is commonly misunderstood .
The answer to Raid's question, "
...if you were given a camera that was set for one year to ISO 200 versus a camera that was set for one year to ISO 2000, which camera would you be using in 2015, and why" is both complicated and important.
With film cameras the recording media's sensitivity can be changed. But all digital cameras operate at a single analog sensitivity level, base ISO. Base ISO is often 100 or 200, depending on the camera's design.
Also, different brands, and different design generations within the same brand, require different strategies when it comes to choosing the optimum ISO. Photographers can’t make an informed decision about ISO selection if they aren’t aware of their camera’s data stream characteristics.
It turns out JPEG photographers and raw photographers should often use different ISO selection strategies.
A simplifying concept to address Raid's question is not to just think only about high ISO noise, or noise as a single parameter in any context. Technical IQ (and the dynamic range - DR) is primarily determined by the analog signal-to-noise ratio, SNR of the data measured by the sensor.
Raid’s question was succinctly answered by Professor Emil Martinec (University of Chicago) in another
Forum. I include an excerpt below.
What is the appropriate mantra? *I would prefer "Maximize Exposure"; maximize subject to three constraints:
(1) maintaining needed DoF, which limits how much you can open up the aperture;
(2) freezing motion, which limits the exposure time;
(3) retaining highlight detail, by not clipping wanted highlight areas in any channel. *
[…]
the prescription is to set the exposure (shutter speed and aperture only) according to (1) and (2); back off the exposure if at base ISO and you are compromising (3). *If you are compromising (3) with your chosen exposure and you are not at base ISO, then you should have started with a lower ISO. *Afterward, depending on the specifics of the camera's noise profile, further optimization results from raising the ISO, up to the limit specified by (3), or the camera's ISO point of diminishing returns, whichever is arrived at first.
Professor Martinec’s advice will always optimize an image’s analog SNR. When time is of the essence, optimal SNR is much less important than spontaneity. Though even then, understanding the disadvantages of gratuitously high ISO settings is useful.
Finally, technical IQ (SNR and DR) is only one aspect of making photograph and it is often not the most important.
Further details are discussed below.
----------------------------------------------
For photographers the signal in SNR is exposure, Exposure determines the total light energy measured by the sensor when the shutter is open. Once the shutter closes exposure (signal) can not be increased. Only two parameters, shutter time and aperture, control exposure. ISO plays no role in exposure.
There is no exposure triangle for digital cameras. Increasing the ISO parameter only amplifies the analog signal levels (and the noise) between the time the shutter closes and when the signals are modeled by the analog-to-digital converter, ADC. This means you can blow highlights two different ways. When the light level exceeds the sensor's maximum analog capacity, the highlights are lost. Or, when the sensor accurately measures all the highlights, increasing ISO can cause signal level to exceed the maximum analog signal capacity of the ADC. In both cases highlight information is lost, but in the latter case the problem is not cause by over exposure, it is caused by over brightening (amplification) the signal levels after the shutter closes.
Shutter time and aperture determine exposure but ISO only affects brightness after the shutter closes. Brightness is not exposure.
The noise is a combination of photon noise (inherent to nature of light) and electronic noise (read noise) from the camera’s data stream. About half the noise in every digital image comes from photon noise (
link). The rest of the noise is generated by the camera's electronics.
In some cameras the noise level due to ISO amplification is much lower than the noise generated by ISO amplification.
In other cameras ISO amplification degrades the SNR. Often the design of the ADC requires higher levels of analog signal amplification. These cameras use multi-stage ISO amplification circuits. Many Canon designs use multi-stage ISO amplification. Another example is the M9 families’ data streams. Both
empirical and
statistical analysis of un-rendered M9 raw data show one should avoid ISO 2500 or greater due to the noise level. Similar conclusions are part of a thread here on RFF (which I could not find using RFF’s Search tool).
Some cameras avoid electronic signal amplification altogether. These designs (Nikon D7000 and others) only use digital multiplication of the raw data file numbers to increase brightness.
Many cameras employ electronic amplification at lower ISOs (below 1600) and a combination of electronic and digital amplification at very high ISO settings.
In this
link Bill Claff compares the DR of many cameras vs ISO. In all cases DR decreases as ISO exceeds base ISO. DR depend directly on SNR. So the decrease in DR is caused by the decrease in exposure recommended by the light meter as ISO increases. If you use base ISO and find the shutter time recommended by the light meter (or your experience) is too long to freeze camera or subject motion, you increase ISO. Now the meter recommends a shorter shutter time. A shorter shutter time means less exposure which reduces the SNR as well as the DR. By selecting different cameras in Bill Claff’s study you can tell which cameras use multi-stage ISO amplification designs. For instance the Leica M240’s data stream properties are very different than the M9 response described above. The Canon EOS 5D Mark II clearly uses multi-stage ISO amplification.
