Metering tonality...18% or 12%?

Over a couple of decades ago, during a photography forum exchange between myself and Ctein (PhD from Caltech, author of Post Exposure, and a reknown color printer), I became familiar with the ISO standard equations for calibration of incident light meters and for calibration of reflected light meters. Ctein managed to convince me back then, that calibration reflectance was somewhere around 12-15% back then for reflected light meters, to match the brightness of the light used for incident calibration. That was after Dick Dickerson of Kodak had convinced him of this as well. Back then computer histograms and digital tools were not as readily available to everyone, but the arguments presented were convincing about calibration.

But calibration itself does not mean that in practice a reflected light meter ought to measure 12% rather 18% for purposes of exposure (not calibration)...after all, we can calibrate thermometers in 33 degree F/0 degree C ice water, but measure 68 degree temperature water in the darkroom or 101 degree fevers. And given the fact that both the reflected light ISO calibration equation and the incident light ISO calibration equation each contained a variable value which was chosen by the manufacturer, it was a somewhat moot proof to try to really conduct! You actually can read about this topic in Wikipedia http://en.wikipedia.or​g/wiki/Light_meter, complete with the equations...much easier than trying to learn about this stuff 20 years ago, much less to prove or disprove it.

And, in more recent times, I had proven to myself that my Minolta Autometer Vf incident meter would render the exact same reading as using my Canon 40D to reflected light meter an 18% gray card. That was proof that 12% calibration is not necessarily the same as using 18% for everyday metering, right?!
Wrong! Let's prove this...

I had photographed 18% gray cards and looked at the histogram, and noticed that the spike was NOT centered in the histogram, even though 18% gray cards were typically valued with R-G-B values of 128-128-128, in the middle between 0-0-0 and 255-255-255. Then someone recently posted on POTN about Lastolite offering a new EzyBalance 12% target and whether 12% or 18% was better for White Balance. That prompted me to think about the old 12% gray card vs. 18% gray card issue for everyday metering, especially since photographing an 18% gray card per the meter did not put it in the center of the histogram!

So I shot A) an 18% gray card per what the 40D meter said (same as Minolta Autometer Vf incident light meter), and
I also shot B) the same 18% gray card per about a 12% gray card tonality metering target.
And here is where the histogram placed the spike for the mid-tone 18% card is shown in these two screen shots...

A.

Interesting, Wilt, certainly bears up a lot of "rule of thumb" conclusions reached by a lot of people!

davidfig wrote in post #12077929
Now for the final nail. Wilt! which one will expose to the left, as 18% or 12% target?

Not sure that I understand what question you are asking!

It does make one wonder if what was behind the need for ETTR was fundamentally caused by the inherent underexposure when using some incident meters or the reflected light meter in the camera!

Wilt wrote in post #12077945
Not sure that I understand what question you are asking!

It does make one wonder if what was behind the need for ETTR was fundamentally caused by the inherent underexposure when using some incident meters or the reflected light meter in the camera!

Opps! sorry I deleted the post.

What I was trying to get at was, should we meter with 12%?

Should we have our exposure at the center or to the right?

Wilt wrote in post #12077744
So this lends credibility to the explanation that the way that Kodak came up with the 18% card was that it was in the middle of the tonal scale -- but 18% was not the target intended to be read by a meter, and also why even Kodak tells you to adjust by about 1/2EV the reading taken with an 18% gray card.

Here's a point I've long wondered about.

I have in front of me the Kodak Darkroom Dataguide, published 40+ years ago. It contains an 18% gray card.

In the instructions for using the 18% card it says:
"When held in the position of the subject and read with a reflected-type meter, it will (on the average) give the correct reading for calculating tungsten exposures. Add 1/2 stop to that given by the meter for daylight exposures."

I don't understand why the difference between tungsten and daylight. I assume (although they don't say) that the tungsten image will be shot using tungsten film since they say nothing about filters.

I don't think I've seen a 1/2 stop difference in DSLR exposures between tungsten and daylight. But I've never conducted tests.

Thoughts?

-js

davidfig wrote in post #12077956
Opps! sorry I deleted the post.

What I was trying to get at was, should we meter with 12%?

Should we have our exposure at the center or to the right?

Based upon my discovery that metering 12% gray card results in correct placement of 18% midtone in the center of the histogram, I very strongly lean to "YES!". It might be that the practice of 12% gray card reading eliminates much of the need to practice ETTR to better capture shadow tonality.

Now if only we could buy the Lastolite Ezybalance 12% target...it is availalble in Europe and Canada, but currently impossible to find in USA.

number six wrote in post #12077957
Here's a point I've long wondered about.

I have in front of me the Kodak Darkroom Dataguide, published 40+ years ago. It contains an 18% gray card.

In the instructions for using the 18% card it says:
"When held in the position of the subject and read with a reflected-type meter, it will (on the average) give the correct reading for calculating tungsten exposures. Add 1/2 stop to that given by the meter for daylight exposures."

I don't understand why the difference between tungsten and daylight. I assume (although they don't say) that the tungsten image will be shot using tungsten film since they say nothing about filters.

I don't think I've seen a 1/2 stop difference in DSLR exposures between tungsten and daylight. But I've never conducted tests.

Thoughts?

-js

maybe more likely that they knew the meters would be affected by the differing light. But then that would also suggest that new meters may have overcome this weakness.

number six wrote in post #12077957
In the instructions for using the 18% card it says:
"When held in the position of the subject and read with a reflected-type meter, it will (on the average) give the correct reading for calculating tungsten exposures. Add 1/2 stop to that given by the meter for daylight exposures."

I don't understand why the difference between tungsten and daylight. I assume (although they don't say) that the tungsten image will be shot using tungsten film since they say nothing about filters.

I don't think I've seen a 1/2 stop difference in DSLR exposures between tungsten and daylight. But I've never conducted tests. Thoughts?

-js

The issue might be due to the different spectral response of tungsten balanced emulsions vs. the spectral response of a meter designed for daylight balanced response. Now that solid state sensors are used for both, that difference is likely to no longer exist in the dSLR.

Wilt wrote in post #12077982
Based upon my discovery that metering 12% gray card results in correct placement of 18% midtone in the center of the histogram, I very strongly lean to "YES!". It might be that the practice of 12% gray card reading eliminates much of the need to practice ETTR to better capture shadow tonality.
.

I agree with you. So I've noticed that Nikon's will expose to the center while Canon's are more likely to expose to the right. Does this mean that Nikon is measuring 18% and Canon are measuring 12% (crap I hope I got that right).

I feel my dyslexia kicking in. Thank god for my indecent meter

So Wilt, when you shoot that gray card (metering for the 18%), how does the in-camera histogram show up -- to the left like the LR one?

Tony, in-camera Histogram shows the peak at the line to the left of center, when exposed per the 40D metering the 18% gray card. The peak is just a little to the left of center, when exposed per 12% target reading. (For this test I put the camera into JPG storage mode.)

So, I'm not sure how much that line represents -- 1/2 stop or more than? I dunno, although if the LR histogram is accurate that meands that even exposing for the 12% leaves the in-camera histogram a bit to the left...which would you trust the most?

I guess if you are getting the right light on the 18% gray card, just bumping to where it hits the mid-point in-camera would be the best thing to do even if it was a bit "to the right"!

tonylong wrote in post #12078791
So, I'm not sure how much that line represents -- 1/2 stop or more than? I dunno, although if the LR histogram is accurate that meands that even exposing for the 12% leaves the in-camera histogram a bit to the left...which would you trust the most?

I guess if you are getting the right light on the 18% gray card, just bumping to where it hits the mid-point in-camera would be the best thing to do even if it was a bit "to the right"!

Should be a bit to the left.

Nobody ever wants to know this, but all of the facts ought to be factored into any realistic arguments...

Any RGB data we can ever see has been gamma encoded. http://en.wikipedia.or​g/wiki/Gamma_c...tanda​rd_gammas

Simple fact. We all know sRGB and Adobe RGB and others are gamma 2.2, and what it means is that ALL RGB data has been gamma encoded. (duh...) Yes, the RAW data is linear at the sensor, but any RGB image anywhere (including any histogram) is gamma encoded.

Middle gray is NOT at midpoint (of anything except the human brain response). 18% is the printed ink color the brain may perceive as middle gray, but we hope to preserve it at 18% at all times, so the brain will perceive middle gray from it. Meanwhile, there are no human eyes involved in digital, and 18% is not 50% of anything.

Light meters do meter for about 12%, which is not about 18% gray cards. Kodak does say to open 1/2 stop from a metered 18% card.

Fact: 128 may be the linear midpoint of 0..255 data, but all RGB data is gamma encoded, which moves 128 (which we mistakenly call histogram midpoint) up to about 187. Easy to see this: An intentional one stop underexposure does NOT move the 255 edge down to be 128. It would, in the linear data at the sensor, but what we see is the gamma encoded histogram.

-1 stop = (0.5 ^ 1/2.2) = 0.73, x255 = 187 73% of 255 full scale

One stop is 50% of the light (midpoint). This 187 varies... It depends ... on white balance and contrast shifts (S curves), and other manipulations, but it is far from 128. 255 shifted down one stop will be much closer to 3/4 than to 1/2. Easy to see. Due to gamma encoding.

18% is 18%, it is not 50%. It is not the middle of anything digital.

After gamma encoding, 18% Gray card = (0.18 ^ 1/2.2) = 0.46, x255 = 117, or 46% of full scale.

46% is coincidentally near middle, but not because either is midpoint of any thing. Midpoint is 187 at 73%, and 18% gamma encoded is 117 at 46%. I'm NOT saying this is a target goal to be sought, but merely that we will often see this if we photograph 18% cards.

The funny thing is people recalibrating their light meter to put a gray card at histogram 128 (calling it "midpoint"). Coincidentally, it is just a minor error, not harmful, but it is just plain dumb. Certainly the wrong idea. 18% is not the midpoint of anything, and actual midpoint is up near 187.

OK, Wayne, got your discertation about Gamma and encoding. But explain, please, for further understanding:

• if there are 256 levels of tonality within the scale of absolute black to absolute white, 128-128-128 in pixel color encoding is the midtone that is like 18% gray.
  
• And that same tone is the standard background used in Lightroom page layouts as seen in the screen shots in Post 1.
  
• And if we use an absolute black at the left and an absolute white at the right end of our camera's histogram (whether the camera has a poor 6EV of dynamic range or (today's state of the art) 8.5EV of dynamic range, what is wrong with that same midtone being exactly in the center? (After all, we merely have fewer divisional lines between the two extremes of black and white, when the dynamic range is less.)