Noise is subjective - its like cholesterol .... there's the good cholesterol and there's the bad
It is even worse; the raw conversion changes the appearance of the noise. For example the "color noise" is not existent in the raw data - the noise is always within a color channel.
Can your favorite pet, RawAnalyze, describe which noise looks like friendly film noise and which noise looks like digital noise?
I have a suggestion to you, Eric. Pls upload raw files with some "good" and with some "bad" noise and we will see if I can "crystallize" the difference on raw level, i.e. before the raw conversion.
I hope we agree on one aspect: in order to judge (and, of course, to quantify, if possible at all) the noise, one needs smooth, unicolored, uniformly illuminated patches. An extreme example for what is useless: you don't want to judge the noise on a multicolored, fine textured area, like a fluffy cloth of mixed color, do you?
As to what I can find in the raw files with Rawnalyze, pls see the
A900 noise analysis. It explains among others the source of blotchiness.
Qualitative assessments are a must, but they don't answer all questions. Moreover, in order to be able to make subjective comparisons, the sample shots have to be done in identical setting. Half stop higher or lower exposure changes the noise in the very deep shadows a lot.