D7000 Full Well

[bjanes]

But I did come across a calculation that deduced the full-well capacity of the 7D pixel is 29,500e-. Whether or not that's accurate, I have no idea.

But for argument's sake, lets assume such simulations and estimates are true. Let's assume that the 29,500e- full-well capacity of the 7D is true. The D7000 with slightly larger photosites and a slightly larger sensor than the 7D, could then reasonably have a full-well capacity in excess of 30,000e-.

Ray,

Your data about the full well of the D7000 appears reasonable. One can extract a wealth of information from the DXO data. Their full SNR graph for the D7000 is shown below. For some reason they express SNR in decibels according to 20*log (mean/std dev); I am not an engineer, but I would think that 10*log(mead/std dev) would be more appropriate since we are dealing with counts and not voltage as Peter Facey notes. Anyway, the SNR at 100% of gray scale for the D7000 is 44.4 dB according to my interpolation, or 166:1.

According to Roger Clark, the square of the SNR equals the number of photons (photo electrons) collected. 166^2 =  27,556. At a SNR of 1, the gray scale value is 0.01%, or 13.3 stops below saturation, corresponding to the DXO reported DR of 13.35 stops.

Regards,

Bill

[Ray]

Ray,

Your data about the full well of the D7000 appears reasonable. One can extract a wealth of information from the DXO data. Their full SNR graph for the D7000 is shown below. For some reason they express SNR in decibels according to 20*log (mean/std dev); I am not an engineer, but I would think that 10*log(mead/std dev) would be more appropriate since we are dealing with counts and not voltage as Peter Facey notes. Anyway, the SNR at 100% of gray scale for the D7000 is 44.4 dB according to my interpolation, or 166:1.

According to Roger Clark, the square of the SNR equals the number of photons (photo electrons) collected. 166^2 =  27,556. At a SNR of 1, the gray scale value is 0.01%, or 13.3 stops below saturation, corresponding to the DXO reported DR of 13.35 stops.

Regards,

Bill

Bill,
It's interesting to see logical and scientific descriptions of the processes involved in such concepts as DR, read noise, shot noise etc. But I never make up my mind until I've seen verification of such theory in practical experiments.

MFDB users often claim that the DXO results are 'way out' because such results do not include the effects of the 'magic sauce' of C1, yet they provide no comparisons with the real contenders, such as the D3X.

You ( I believe), BJL and Erik seem convinced that the extra couple of stops of DR claimed in the DXO results for the D7000 is irrelevant because it's drowned in shot noise.

I don't believe it till I see it. I'm hoping I don't have to order a D7000 just to see for myself, comparing DR performance with my 50D, then cursing and swearing and stubbing my toe if it proves to be no better in the shadows than my 50D.

I also sense a certain confusion about absolute (or engineering) DR, and useable DR.

It may well be the case, in fact I'm sure it is the case, that the 14th and 13th stops of DR in the D7000 are useless, because of both shot noise and read noise.

But what about the 11th and 12th stops of DR? That's the question.

I know that the 11th and 10th stops of DR on my 5D are useless, because I took the trouble to find out. I didn't need to refer to Roger Clarke.

I suspect the 11th and 10th stops of DR on the cropped format D7000 would be much more useful than those same stops on the full frame 5D, or the new 60D. If so, that's worthwhile progress in my view, and reason to buy the camera.

[thierrylegros396]

So how does compare good old Fuji S5 Pro relative to D7000 and MF ?!

[Ray]

So how does compare good old Fuji S5 Pro relative to D7000 and MF ?!

Thierry,
DXOMark would suggest that DR is brilliant up to and including ISO 800, for the S5 and S3 Pro, but then takes a dive as steep as a high diver in a swimming pool.

I guess the problem is, although these cameras have 12m separate pixels, they don't deliver 12mp resolution. They are essentially 6mp cameras (maybe a bit more).

Not only is performance above ISO 800 very disappointing, SNR, tonal range and color sensitivity seem to be seriously lacking across the entire ISO range, according to DXOMark.

[joofa]

For some reason they express SNR in decibels according to 20*log (mean/std dev); I am not an engineer, but I would think that 10*log(mead/std dev) would be more appropriate since we are dealing with counts and not voltage

In engineering the factor 20 is used when you are dealing with direct signal values (which includes voltage) and 10 when dealing with its square (power).  Since, mean/std_dev is a ratio of signals, hence, 20*log is the right formula to use .

Joofa