Dynamic Range of CMOS

[davidvj]

Just what is the dynamic range of the CMOS in my Rebel XT?
This was something that with B&W film I could really control and I am assuming that the Canon CMOS is providing a far better range than for instance Kodachrome .. but I do not appear to find anything in the specs to confirm this.
Is metering still assuming an 18% gray as the mid point?
Is there a 'toe' and 'shoulder' to the data capture process or are we working on a linear capture?
Is there some place that all such information is published?

[Quad]

I have been wondering also. Reading tells me it is linear with about a 6 maybe 7 stop range.
I just ran some simple tests. I exposed a grey card from zone minus 1 to zone IX in RAW. I put a color sampler in the middle of the frame in CS2's RAW converter and adjusted the exposure (and brightness at the extremes) so that the sampler read 128 (middle grey).
It looks linear from Zone 0 to zone VI Zone zero took a +4 exposure increase and brightness went from the norm at 50 to 100.

Zone minus 1 raised the brightness to 150 but the noise was high but I think that looks linear as well. Brightness of 50 seems to be one stop.

Zone VII was decreased by 1.7 (not the expected 2 if it was to be linear) Zone VIII 2.25 (not 3) so that looks like a bit of a shoulder. Zone VIII and a half would not correct at - 4 stops (it was 147 instead of 128 ].

At Zone VIII photoshop was giving me highlight warnings before the correction.

Zone IX did not result in highlights that could not be recovered BTW. A minus 1.75 brought them in line so there appears to be lots of room to recover blown highlights but I did not shoot above Zone IX to confirm the extreme here. (I have to leave something for the next rainy day)
I am not sure of my methodology as I just made it up and I am very new to all this digital stuff.

What do I (a rank amateur at this) conclude. We have quite a bit of lee way in the shadows and can photoshop a very nice shoulder, it looks like. The straight line in a film goes from about Zone III to Zone VII (Adams said IV to VIII but he just reads it different than I do if I look at his graphs). But as he says that is where to most detail is anyway. We do have a bit of less lee way in the highlight portion of the graph by at least a half stop.

Is this coherent to anyone but my four multiple selves? Is my methodology bunk?

[Tee Why]

I hear that it's similar to a slide film but less than film for prints. I keep hearing 5 stops.

[Duder]

I've been hearing 5-7 stops for a while now, although I recently read in a magazine which tested the 10D sensor's DR and found it to have a 9 stop range.

[Robert_Lay]

The really determining factor is whether you shoot RAW or JPG.

The things that you can do in RAW processing so far outpace what you get with JPG that they cannot even be compared.

In regard to dynamic range, using JPG, I have made careful measurements on my G5 and it is at least 9 stops. However, that is not linear - it's about like a film response but with a much sharper toe and a shoulder a little sharper than film.

The myth of 5 stops has an unbelievable origin. There are many who seem to think that since the histogram display on the cameras is divided into 5 approximately even regions, that that is the dynamic range. Nothing could be further from the truth.

In regard to the amount of overexposure leeway in RAW, I have made very careful measurements of that on the G5 and found it to be nominally 1 f-stop.

The second of the two tutorials listed below has the later details on my experiments with "exposing to the right" and the use of the clipping display in the histogram and related RAW processing procedures.

Tutorial on RAW Processing as a downloadable PDF:
http://www.zaffora.com/W9DMK/RAWProcessing.pdf

Quick Tour of RAW Processing
as downloadable PDF version:
http://www.zaffora.com/W9DMK/QuickTourOfRAWProcessing.pdf