selmslie wrote:
We [should] already know that visible noise (signal to noise ratio aka S/N or SNR) is the determined by exposure, not ISO.
I have studied this at length and determined that S/N is primarily the result of shot noise. Sensor read noise does not begin to influence the SNR until well below middle gray (around EC-5 or darker).
If we develop an image from raw, there are times when you find that the original exposure may have resulted in a dark JPEG from the camera.
Increasing the ISO in the camera or reducing it and using the Exposure slider to offset it might produce the same image, assuming the camera is ISO invariant, which nearly all modern cameras are (but that's a different topic).
What happens to noise when you move the Exposure slider to the right to brighten the image? With each stop added via the Exposure slider we increase the apparent brightness by one stop. But we also double the effect of shot noise and this lowers the visible SNR.
So what would happen if we took two images using the same exposure (aperture and shutter speed) and equalized the brightness with the Exposure slider?
It has been suggested that the image that used the higher ISO would have less noise. Of course, we already know it would be better because it will be closer to ETTR (exposing to the right) but that may actually be because the shadows are recorded better.
To determine if there is really a measurable difference, I tested three cameras, a Fuji X100t, a Sony A7 II and III and a Nikon Z7. As luck would have it, all three are ISO invariant.
Below are the results of the test.
The samples for each camera used the same exposure (aperture/shutter speed). The SD column is from a 150x100 pixel selection at the center of the image.
The A7 II has not Bayer array so the effective ISO settings are actually one stop higher than what was set. ISO 25600 does not work right but all of the other ISO settings in the analog range are proportional and ISO invariant. The log(SD) changed by almost exactly one stop with each doubling of the ISO. For the other cameras I just tested two ISO settings.
As we can see, there may be a very small difference in the noise level changes but they are within a reasonable range of the actual difference in ISO.
If you have been following this so far, it conclusively shows that raising the ISO does not actually alter the amount of visible noise. It's safe to say that there is no benefit to lowering the ISO and using the Exposure slider to correct the apparent underexposure.
The higher exposure does not suppress any noise. In fact, with an ISO invariant sensor, it has absolutely no effect.
ETTR remains the recommended approach because it collects more clean information in the shadows, if you decide to amplify it.
We should already know that visible noise (signa... (
show quote)
I have found that Topaz DeNoise and even Lightroom's Denoise function do wonders for unwanted image noise...and that's even without all the dialog and charts...
I'm sure other programs will equally deal with unwanted image noise.