I assume that unity gain is at ISO 400 for your camera.
Hi Bill,
It is (as
determined separately), but that's not exactly what the chart based on Jim's method is showing. That chart sort of shows
the ratio of Photon shot noise to actual SNR a few stops below mid exposure level (say 5 stops below saturation).
Bill Claff's tool indicates that an additional half stop of photographic range can be obtained by increasing the ISO from 400 to 1600 as shown below. Thus, unity gain does not seem to be the final arbiter in determining when raising the ISO ceases to improve DR.
I have a bit of an issue with a dynamic range criterion for underexposed images. Engineering Dynamic Range at native ISO is a useful indicator about how deep into the shadows a sensor is able to discern detail in an ETTR shot at native ISO gain. That tells something about noise floor, and quantization and well depth (when we determine the saturation level as well). Bill Claff's 'photographic dynamic range' (PDR) uses a higher, slightly arbitrary, SNR of 20 as the lower cut-off point. So his charts basically show the SNR improvement at that specific level as a function of ISO gain setting, expressed as a PDR number.
However, as useful an indicator for shadow detail that may be, that says little about how good the higher exposure levels will render. As my quick test of a ColorChecker shows, patches 1-6 are CC gray level patches 19-24, boosting the ISO above Unity gain levels also boosts noise levels of brighter areas (more than of the shadows!) in the image
after Raw conversion:
ISO 800 gain +1 stop push in postprocessing is overall better than ISO 1600 gain, even ISO 400 gain +2 stop push is better than ISO 1600 gain.
Bill's photographic DR has a noise floor well above 0 dB, so the difference in engineering DRs might be even higher. As I understand the situation, increasing ISO above unity gain will not improve quantization, but if the read noise continues to decrease beyond unity gain, there will be an improvement in DR as the read noise decreases. Read noise for the Canon 1Dm3 is shown in Figure 15a of Emil's treatise. The asymptote of Emil's graph agrees with Bill Claff's tool for this camera.
What do you think?
While it is useful to improve the quality of the deepest shadow detail, by boosting the ISO gain too much we will simultaneously increase the noise level of somewhat brighter scene areas. If there is more relevant detail in the brighter scene regions, we may need to decide to be a bit more modest in boosting the ISO gain all that far. Afterall, it's in those low light situations that specular highlights and direct lightsources can start to become an integral part of the image content/composition. When I e.g. shoot church interiors, or other architecture and outdoor night scenes with available light, the color of the lightsources (candles / incandescent bulbs / traffic or street lights / specular material surface reflections) are something I do not want to clip. Generating too much noise in brighter regions that needs noise reduction may also hurt detail that would be easier to retain with less need for noise reduction.
So the short version is, image quality may sometimes benefit more from not pushing ISO gain to the extreme, even if we may lose a fraction of a 1/3rd stop DR to theoretical 'under-amplification' of the noise floor.
Of course, also the Raw converter plays a role in this. Using a linear tone curve will retain highlight noise more than a heavy highlight roll-off such as the default Process 2012 in ACR/LR.
Cheers,
Bart
