Advanced Bracket Features Request

[dmilligan]

Ah yes, I remember reading bits of that conversation, that guy had no idea what he was talking about. I think he was missing the simple fact that x / sqrt(x) = sqrt(x). So if shot noise is the std dev (σ) of photon distribution and σ = sqrt(N) since it's a poisson distribution, which means the Signal/Noise = N / sqrt(N) = sqrt(N). It's a funny property of photon shot noise that the SNR also equals the amount of noise which also equals the σ of the signal. All because x/sqrt(x) = sqrt(x). The equations weren't wrong, he just didn't realize that they were mathematically the same thing.

[Audionut]

He was just trolling.

I'm pretty sure in your post on the last page, you have mixed SNR with bit precision.  An 8 bit display has 8 bits of precision.  In other words, there are 256 possible shades of grey.  Here, a captured tonal precision of greater then 256 can actually causes problems, when this increased tonal precision is mapped down to 256 levels (banding).  The standard practise is to dither this information (add noise) to mask the levels and reduce the effects of banding.

Stretching and compressing is all in the tonal range, and separate to the SNR of each tonal step.  When we get down to -6EV below sensor saturation (14bit capture device), each successive tonal step below this now has less precision then a typical display (I say typical now because hopefully typical displays are now truly 8bit and don't simply dither down further to 7 or 6 bits).  And of course, this is only a problem where you a stretching the shadows. -6EV being below the recommended 4EV spacing for togs, and extremely below all of this 0.3EV spacing nonsense.

SNR is different entirely, and of course, is a ratio of signal vs noise.

To put things into perspective for those of us less mathematically inclined (3+3=8 right!), -1EV below sensor saturation is still considered to be white.  Not a highlight, white.  The top 2-3 zones in the Zone System.