Bart,
Thank you very much for your detailed investigations. I repeated my investigation using Rawdigger to separate the green channels from two images (Image 15 and Image 16) of flat fields taken under identical conditions and exposure with the Nikon D800e.
Hi Bill,
Thanks for sharing your findings, it makes me feel not too lonely
I haven't had the time to compare Rawdigger channel output with the others yet, but its TIF compression is complicating matters slightly.
For completeness, I also performed subtractions using ImageJ using non-cropped files, since I don't know how to crop by the numbers with ImageJ--it seems to require a manual selection. The results are nearly identical to those obtained with cropped images.
It's simple, once you know it ... For a central crop, just use "Image>Adjust>Canvas size... Position: Center" and set the crop dimensions. Since we're shrinking the canvas, the Zero Fill option is not important.
As you found, subtracting G2 from G1 of a single image gives a larger standard deviation than subtracting either the G1 or G2 from two identical exposures.
Yes, and that is why it's always more accurate to use exposure pairs, because then we sample with the same sensels, thus eliminating one variable.
Looking at the subtraction of the green channels of a single file, I see some banding which is confirmed with FFT analysis.
[...]
No such banding is noted when subtracting crops from two separate files.
Indeed, and that is not uncommon. In fact that also demonstrates the superiority of the 2-frame subtraction method. Even with 2 exposures most of the pattern noise is eliminated, only random noise remains. Even a Fourier transform doesn't pick-up the parallel readout / crosstalk signal anymore.
That's not to say that the G1-G2 subtraction has no merit. In fact, it can be used as a differencing tool to amplify some correlated signals with two offset windows, but it still does reduce the Standard deviation so some noise is reduced, and it isn't the truly Random signal.
My investigations are consistent with your theory that PRNU is not eliminated when one subtracts two green channels from the same file. For sensor analysis, one should use two images obtained under identical conditions as standard practice. The ability of Rawdigger to separate out the channels does offer an additional tool for image analysis.
Well, some PRNU may be reduced, but other signals may be amplified (e.g. hot/dead pixels). That's the remaining puzzle, to find out what it is best used for. As I recall, Emile Martinec also used the G1/G2 for some of his noise analysis, when lacking the benefit of an exposure pair. I'll have to look it up and read it again, with the benefit of the current insights.
Anyway, nothing beats the old tried, tested, and proven subtracted-exposure-pair technique, but there may be uses for the single image G2-G1 evaluation. It e.g. detects dead/hot pixels which therefore points out that an outlier will spoil the statistical analysis if allowed to play a role in the analysis of that area.
Cheers,
Bart
