See now there you go... I made a simple statement of fact that SNR goes up with signal level and that is too simplistic and you want to get into "image quality" which I specifically wanted to avoid because here we go. I call a NR reduction algorithm "clever mathematics" but you want to say it's not "clever mathematics"... whether it's clever or not clever is a matter of opinion I guess but I think it's at least a little bit clever. Can you give me that one?
You're obviously knowledgeable about this topic. Me too. Should we compare "credentials"?
Honestly, I don't want to offend you or anyone else. Someone was offended because I got smited... YES! Probably for starting this so I may deserve it
Many of your points are technically sound. Sure, counting photons in 4 bins and adding vs. counting in 1 bin is the same total number of photons. But there will be less signal (fill factor <1) and more shot noise and read noise and therefore lower SNR and DR and you've agreed the accuracy of counting those photons is not as accurate if only as a technical matter. If the SNR is high enough across the entire image then the whole argument is moot but if not you can go ahead and apply a NR algorithm to try and improve the image quality.
I do agree that the more data you have to work with the better NR algorithms will work. The question would be where is the sweetspot above which the additional noise due to smaller pixels outpaces the improved application of the NR algorithm.
And why is the DR in the Nikon D7000 16MP sensor (Sony) so high? If you look up the full-well capacity on sensorgen.info (ooops, that could be another debate) you will see it's very high... probably because they improved the fill factor (i.e. a larger photosite in each pixel, the size of the photosite matters). They also got the noise down in that sensor. Result = large DR of 14. Now look at the Sony A77 (24MP) presumably using at least the same tech as the D7000 sensor and you'll see the full-well is much lower and therefore the DR is only 12.2 so the evidence suggests again that photosite size matters at least to the technical characteristics (again I'm trying to leave "image quality" alone).
Anyway, I apologize to everyone, their families, their friends, and the whole world for starting this.
-1 to myself