The normalization is about the whole image, but the screen s/n is about pixel to pixel comparison.
So the K3 vs the D800e is close to a stop worse at iso 3200. The K5iis is pretty close to the same. Pixel to pixel.
Same lens, same subject distance and cropped to show the same scene would have less pixels on the K1 vs the K3, and the perceived noise would be what?
If I could get the same scene framed into either formats by moving or lens change, the full frame has a substantial improvement.
I am not certain for my shooting, and am trying to work out the difference. I suspect that for some circumstances it would be better, some worse, and more than likely different choices would be made to get the best shot in any circumstance. That is what I'm trying to figure out.
So far, this is what I know.
Depth of field is immaterial. If I get it in focus on my K3, I will have it in focus on the K1. That is my primary concern; no giving up a stop necessary.
At pixel to pixel comparison, the K1 is close to a stop better noise characteristics at higher isos.
The pixel density is lower, about where the K5iis was.
I would like to see a chart similar to the print noise s/n where the calculation has an iso setting, a % grey and % crop from the sensor size for both the K3 and K1. That is what I am interested in. A normalized s/n at various crops.
---------- Post added 03-17-16 at 10:20 AM ----------
Detailed computation of DxOMark Sensor normalization - DxOMark
The formulas they use are there. I have to go do some work to make money to afford this stuff, but intend to try to plot something.