does R 128, G 128, B 128 = Zone 5?

[gaetano catelli]

does R 128, G 128, B 128 = Zone 5?

 

if so, what would be the RGB values for Zone 6?

[kevinteo]

Yes. And Zone 6 = R 153, G 153, B 153.

[gaetano catelli]

thanks. is there a link to a chart with the whole shebang, or an algebraic formula that you know off-hand?

[dave_redmann]

I have looked into this, and the short answer is no. There is a formula. I think it depends on the gamma. Frankly I figured it would be simple but it isn't, or at least I haven't figured it out yet.

 

For those questioning, try this experiment: Take a raw file, and do three converstions, one at -1 EV (or stop or whatever your raw converter calls it), one at 0 EV, and one at +1 EV. Pick three points of varying brightness around the image, and note the exact coordinates (so you can be sure to examine the exact same pixels in each version). Use the color picker to get the R, G, and B values of each pixel from each version of the image. If Zones I through XI were represented linearly by 0 through 255, you would find that a change in 1 EV always meant a change of +/- 25.5. (If this were the case, Zone I would be 0, Zone II 26, Zone III 51, Zone IV 77, Zone V 102, etc.) But it just doesn't work that way.

 

There is a useful post somewhere on photo.net addressing the issue. Unfortunately I can't put my fingers on it at just this moment.

[kevinteo]

Shouldn't the scale run from 0 to X? That would mean Zone 0 is 0 and Zone I is 26.

[godfrey]

<center>

<img src="http://homepage.mac.com/godders/zone-system-pixel-values.jpg"><br>

</center><br>

Above is what I've been using as a rough estimator for Zone I to X pixel values.

<br><br>

Godfrey

<br>

[gaetano catelli]

it looks to me like Godfrey's chart indicates that every Zone is about +/-30 relative to the Zone next to it.

 

this would not be surprising since, iirc, one of the distinctive characterisitcs of digital sensor output is that it is a straight line relative to the amount of light falling on it. (i hope i said this correctly).

[gaetano catelli]

make that +/-28.3 (ie, 255/9)

[steven_clark]

Actually the formula would involve not only the gamma, but the assumed dynamic range of the image so you know which zones "white" and "black" fall in since the zone system is logarithmic it has no 0 only infinite shades of "half as much light".

[gaetano catelli]

cancel what i said about +/-28.3

<p>

here are the RGB values i got via an empircal test:

<p>

Zone 0 -- 1<br>

Zone I -- 7<br>

Zone II -- 25<br>

Zone III -- 46<br>

Zone IV -- 76<br>

Zone V -- 121<br>

Zone VI -- 176<br>

Zone VII -- 221<br>

Zone VIII -- 251<br>

Zone IX -- 255<br>

[dave_redmann]

Godfrey, could I direct your attention to the image that you posted? I realize it depends a bit on the monitor, but surely you would agree that the difference between your I and II is <B>much</B> less than the difference between your IX and X? In other words, your scale is not divided into even zones, each representing a 1 EV change. I submit that it is strong evidence that JPEG encoding does <I>not</I> have a linear representation of zones (which are themselves a logarithmic representation of density, IIRC).

[gaetano catelli]

my method was to shoot a grey card 10 times (in RAW), changing the exposure 1 stop each time.

 

i then opened each RAW file in Photoshop, and used the "Info" box to determine the RGB values for the same set of X,Y coordinates in each image.

[dave_redmann]

Gaetano, that sounds like a reasonable approach, at least assuming your raw conversion process is neutral. The only thing I would point out is that, at the very extremes, it is hard to know exactly how much exposure it takes to get pure black and pure white. I think you are on the right track.

 

I plan to try something similar when I get a chance, in addition to more raw conversions of the same file with different density compensation (I've already done some).

[godfrey]

Dave,

 

The posterization technique I used to create 10 evenly spaced zones are graduated linearly

from black to white. It's not an EV step wedge. The differentials between steps are exactly,

mathematically even but our *perception* of the differences differs because human vision

is not linear in nature.

 

This has nothing to do with JPEG compression encoding, which has plenty of resolution to

represent a grayscale stepwedge accurately.

 

I have a Kodak Professional Photoguide with an accurate Zone V card (18% gray

reflectance) and a step wedge registering eight EV steps (80%, 40%, 20%, 10%, 5%, 2.5%,

1.3% and 0.7% reflectance). Since these are known-accurate reflectance values in EV series,

I'll capture an image of them with both scanner and DSLR to render a similar step wedge in

these terms. I don't find it as useful for my work in evaluating digital exposure, however.

 

I doubt I'll get to it until next week, though, as I'm preparing for a trip this weekend.

 

Godfrey