GH3 = Mega PDR test (all ISOs, new results, D800 lost its crown, The Holy Grail is found............

Right.

DxOMark's Tonal Range seems more complicated than I had previously imagined. I put a fair amount of time into deciphering it from what I was able to find, and authored this post here:

http://forums.dpreview.com/forums/post/50118190

... which resulted in a complete void of interest or response. Check it out (if you might).

Since the choice affects how much said "DR" metric is influenced by Read Noise and/or Photon Shot Noise, the choice of the noise-floor does not in my estimation seem arbitrary ...

Using a SNR of 26 dB (a linear ratio equal to 20) as Bill Claff does leads to results which are by far more influenced by Photon Shot Noise than by Read/Dark Noise. Thus, it seems to me to be much more indicative of Quantum Efficiency and Full Well Capacity than Read/Dark Noise.

However, since the test results do not seem to be compared based upon a measure of "Saturation ISO Sensitivity", but are instead based upon JPG-referenced, manufacturer-rated ISO (which may itself deviate from SOS when using REI), the test results seem (to me) to rather disassociated from Full Well Capacity (per se), and are only loosely (and somewhat unpredicatbly) related to Full Well Capacity.

For instance, the non-linear tone-curve transfer-function of the E-M5 (in "Gradation Normal" mode) has an around 4.0 EV input range above sRGB output of 118 out of 255. The E-M5's "Gradation Auto" mode increases that input range to around 5.0 EV above sRGB output of 118 out of 255.

Thus, (in the E-M5, and other cameras, that employ non-linear tone-curve transfer functions at higher input levels), how the JPG-referenced, manufacturer-rated ISO relates to the "Saturation ISO Sensitivity" (and Full Well capacity) Seems to me to not represent a consistent or a straightforward relationship.

Wouldn't you say that using higher reference noise-floors for DR metrics does indeed bring Photon Shot Noise levels into play, and using manufacturer-rated ISO Sensitivities instead of Saturation ISO Sensitivities seem to introduce uncertainties as to how to interpret such test results ?

Reference information below:

ISO SATURATION SPEED is the S value when the exposure level generates a picture with image highlights that are just below the maximum possible (saturation) camera signal value. The adequate average exposure Hm is regarded as 1/7.8 of the exposure level at the saturation point (saturation exposure), where 7.8 is the ratio of a theoretical 141% reflectance (which is assumed to give the saturation exposure with 41% additional headroom, which corresponds to 1/2 “stop” of the headroom equal to an 18% reflectance (the standard reflectance of photographic subjects).

Thus, Equation B.2 can be changed into Ssat = 78/H where Hsat is the saturation exposure in lux-s. The saturation speed only shows the saturation exposure as a result.

Suppose there are DSCs that have the same sensitivity at low-to-medium exposure levels, meaning that their tone curves at those levels are identical. If a tone curve of one of them has a deeper knee characteristic near the saturation exposure, the saturation speed of that DSC becomes lower. Thus, it is preferable to use the saturation speed to indicate the camera’s overexposure latitude.

SOS is the S value when the exposure generates a picture of “medium” output level corresponding to 0.461 times the maximum output level (digital value of 118 in an 8-bit system). Hm in Equation B.1 corresponds to the exposure that produces 0.461 times the maximum output level. The numeric 0.461 corresponds to the relative output level on the s-RGB gamma curve for the 18% standard reflectance of photographic subjects.

SOS gives an acceptable exposure because the average output level of the picture becomes “medium.” Thus, it is convenient for a camera set. However, there is no guarantee that the exposure indicated by SOS is the best. Also, it is not suitable for an image sensor, whose output characteristic is linear.

REI is the S value when the exposure generates a picture with an “adequate” output level that a camera vendor recommends arbitrarily. Accordng to this definition, it is apparent that REI can apply only to a camera set and that the exposure indicated by REI would be adequate only if the vendor’s recommendation is appropriate.

Source: Appendix B, Sensitivity and ISO Indication of an Imaging System, Hideaki Yoshida, Pages 319-321, Image Sensors and Signal Processing for Digital Still Cameras, 2006, Taylor and Francis Group, LLC. http://www.scribd.com/doc/91224062/Image-Sensors-and-Signal-Proces