ISO originally rated the light sensitivity of film and was related to the size of silver halide gain in the film. The larger the grains, the higher the ISO. High ISO films were termed "fast". All of this changed as digital sensors replaced film, but we still needed a way to rate a sensor's reaction to light. ISO was handy and understood by photographers so it was adopted. Only digital ISO has almost nothing to do with the light sensitivity of a camera sensor. A better description of digital ISO is amplification of the signal. That's why noise can occur at high ISO. Easy, so far...
Now comes the concept of ISO invariance. A simple explanation is that increasing exposure in post processing has the same effect as raising ISO in camera regarding noise. Cameras that are said to be ISO invariant do not exhibit noise when an image is underexposed in camera with exposure increased during post processing. There are two caveats: this relates to raw images and the amount of increased exposure is typically limited to two stops. With me so far? Super!
Techo Geek Speak aside, what does any of this mean? Imagine you have a scene with very bright exposure - a musician performing illuminated by a very bright stage light. You have two choices: expose to please the meter and risk blowing out the highlights (remember: they can not be recovered) or underexpose and "correct" in post processing. I chose to underexpose by 1 stop (-1EV) and this was the resulting raw image right out of the camera:
Great huh? Easy, it was what I expected. I used a Nikon D4 with a known ISO invariance (your results may vary, so test this before you make a critical photograph), a 200mm lens wide open at f/2.8 and ISO 4500. Now comes post processing.
In post the exposure was increased +1.5 stops, based on facial exposure. A radial filter was applied around the musician's face to tone down the rest of the scene with a - 2/3 stop slider. The final result after cropping:
See ya next week,