Feedback Please: The Kodak 18% Grey Card and Metering, a new look.

[Chan Tran]

valuable info;I started my evaluation of average street/landscape shots,and so far. I get an average luminance value of 112,which equals abot Zone 5.1 and 17.7 % reflectance;I think,I'm on to something here.

Great work Ralph! So for C=250 and K=14 the reflectance is 17.59% very close to your finding. By the way 112 is the pixel value?

 

[Chan Tran]

This resource propose some calculation and arrives to the conclusion that "K=14.32 is the value that “matches” 18 % reflectance".
So lightmeters such as Minolta and Pentax calibrated to K = 14 are actually calibrated to a reflectance of 18% not 14% and not 8% according to this source.
It generally makes much more sense to me that my lightmeter is calibrated to the perceveid middle grey of the human eye, 18%.

http://dpanswers.com/content/tech_kfactor.php

Like in the post above the incident meter is calibrated to a C=250 and the reflected meter K=14 then the reflectance is 17.59% which really agrees to the 18%.

 

[Diapositivo]

Like in the post above the incident meter is calibrated to a C=250 and the reflected meter K=14 then the reflectance is 17.59% which really agrees to the 18%.

According to that table, the difference between 17.59% and 18% would equal to a difference in exposure of 0.02 EV.
That leads me to believe that Minolta, Pentax etc. actually use a K = 14.32 and declare K to be 14 just because they assume they deal with normal people

 

[Stephen Benskin]

I think we should be careful when using the K and C ratio for average calibration reflectance when moving beyond the "ideal" K and C numbers in the standard. I am putting something together that may address this. This is a difficult subject as some of the finer points appear to be inside the manufacturer kin of stuff.

The math is saying there's nuance, but there's also this thought to consider. A reflected exposure meter has a K of 12.5. An incident meter with the dome has a C of 320. The average reflectance based on this K and C ratio is 12%. When using the same incident with the disk having a C of 250 the average reflectance is 15.7%. Nothing else has changed. The scene illuminance hasn't changed and the reflected exposure meter hasn't changed. Is it suddenly reading 15.7% reflectance? The dome and flat disk are also used differently. We need to be careful of any confirmation bias.

 

[DREW WILEY]

I thought it was once common knowledge at one point that conventional gray cards were 18%, while Kodak's were slightly different, around 12%. But the quality control was so bad that any of these could be plus or minus five percent of alleged value. That's what I learned running stacks of them through a spectrophotometer. For any serious film testing, now I only use a clean unfaded MacBeath gray scale. They do it right. Now you've good
equivalent items being marketed for setting "gray balance" and "white balance" in digital cameras, and some of the more expensive ones are pretty
damn good, matching the 18% of the mid-range MacBeath gray almost exactly.

 

[Stephen Benskin]

Does someone here has the ISO standard. I can't afford to purchase it from ISO. If so can someone post a quote from the standard about the reflectance standard of 12%?

For General-Purpose Photographic Exposure Meters (Photoelectric Type) ANSI PH3.49-1971. In two parts. This is the standard with the good appendixes.

 

[RalphLambrecht]

Great work Ralph! So for C=250 and K=14 the reflectance is 17.59% very close to your finding. By the way 112 is the pixel value?

thanks;No,112 was the average luminocity value,according to the PShistogram statistics.

 

[DREW WILEY]

Reflectance is also inherently related to sheen, so unless that is itself standardized, all bets are off. And in this respect alone, gray cards vary.

 

[pentaxuser]

Can somebody(ies) sum up in relatively simple terms that the bulk of APUGers will understand, what RobC has said that will mislead a user of his article to sufficiently wrongly expose a negative? Isn't this the only crucial way to look at what RobC is trying to do for those who wish to get the right or as right as makes no difference the exposure?

I got the impression that he is trying to help photographers in general to achieve better negatives.

So where has he failed to do this?

Thanks

pentaxuser

 

[Diapositivo]

The problem I see

Can somebody(ies) sum up in relatively simple terms that the bulk of APUGers will understand, what RobC has said that will mislead a user of his article to sufficiently wrongly expose a negative? Isn't this the only crucial way to look at what RobC is trying to do for those who wish to get the right or as right as makes no difference the exposure?

The problem I have with RobC reasoning is linking the exposure value that the reflected light meter would give on the grey card, i.e. "the middle grey", and which is always the same shade of grey, with the "average tone of the scene", which can vary with the situations.
Rob takes into account several scenes with several ranges, and, taking fixed the upper value, he reasons that scenes of different range will have a different average grey, so that the average grey of the scene will be darker, the greater the range of the scene.
That leads him, if I get him right, to the conclusion that the measure indicated by the incident light meter or the reflected light meter on the grey card, based on 18% grey, would not necessarily fit your scene, if your scene has an average grey of, say, 8%.

I might have misunderstood the thought of Rob here, his thought is not yet clear to me (I expressed these thought in a private message but my reliefs were not discussed in the thread, nor in private).

I reason that the reflected reading on the 18% grey card, simply gives you the placement of the 18% grey (the exposure value that will give, on your slide, an 18% grey subject). Examination of the range of the scene, or of the average tone of the scene, is something that the photographer makes as a second logic step. Knowing the placing of 18%, and knowing his scene, he will decide the exposure.

I also reason that the variables existing in B&W regarding exposure, development, and printing should stay aside. Light meters work with slide film, which can basically be processed only in one way, and for which there is a well-known "18% grey" point on the density curve (probably a point of 72% opacity, more or less). That's where the reading of the light meter will be placed.
All the reasoning relating to B&W film, where is base, where is the speed point etc. tend to complicate things unnecessarily. Mesuring the 18% grey card will give you a slide with a "correct" rendering of an 18% grey card, end of story.

A separate problem is what exactly is this middle grey the light meters will give on your slide film: is it 18%, 14%, or 12%, or what. But that should not have anything to do with RobC reasoning.

 

[Chan Tran]

Diapositivo!
I found that if you point the spot meter at something (neutral color) and use that reading for the camera setting for most slide film it would give you a 1.00 D on the film. This can be done by looking at the characteristic curves published by film manufacturers.

 

[cowanw]

The problem I see


The problem I have with RobC reasoning is linking the exposure value that the reflected light meter would give on the grey card, i.e. "the middle grey", and which is always the same shade of grey, with the "average tone of the scene", which can vary with the situations.
Rob takes into account several scenes with several ranges, and, taking fixed the upper value, he reasons that scenes of different range will have a different average grey, so that the average grey of the scene will be darker, the greater the range of the scene.
That leads him, if I get him right, to the conclusion that the measure indicated by the incident light meter or the reflected light meter on the grey card, based on 18% grey, would not necessarily fit your scene, if your scene has an average grey of, say, 8%.

Diapositivo
I have struggled with this very thought for a long time. But I have not had the courage as you have to express myself for fear of getting a ton of coal or snow (as may be from either end of the scale) poured on me. Having you express it gives me courage.
Key, I think, is " taking fixed the upper value" which is of course never fixed. 100% luminance in Santorini at noon is very different from 100% luminance in my home town on the foggy day yesterday.
Perhaps the wording "18% grey card" or "middle grey" is inappropriate for it's use; I think of that as just a Proper name. Another name might be "this is a tone of card that I can use to represent a tone in my finished print".
I thought of a mind game last night while I was unable to sleep pondering this. I cut a long 21 step Stouffer Step Wedge in thirds taped them all up on a window with clear north sky behind and photographed each one. middle grey would surely be different for each but the exposure must surely be the same. (? presumably based on a spot reading of step 10 of 11)

Anyway, like Diapositivo, this stage of the argument is where I get stuck. (which is like the first paragraph of the argument, so I don't get far)

 

[Diapositivo]

Diapositivo!
I found that if you point the spot meter at something (neutral color) and use that reading for the camera setting for most slide film it would give you a 1.00 D on the film. This can be done by looking at the characteristic curves published by film manufacturers.

The curves are logarithmic and I expected "middle grey" to be more or less at the middle of the density curve. More or less. I understand a film can have more range toward the highlights or toward the shadows around middle grey. So "middle grey" is probably not middle of the curve.

Looking at a graph of let's say Fujifilm Prova 100F I expected middle grey to be around 1.5 of density.

Your finding that the grey card gives a density of 1 places the grey card quite a bit on the highlights (less than 2 EV before the curve looses linearity).

The usual behaviour in my Minolta Spotmeter thinking of a slide is that I have 2.3 EV above grey and 2.7 EV below grey. So the central 5 EV of range should be centered somewhere 2.3 EV of the "clipping point" so to speak. That corresponds to around 1.1 or something like that.

So your observation is right and the reasoning I do regarding where shoud I find middle grey on the graph is wrong, I presume. And "middle grey" is not at the middle of the density curve! That might have something to do with the way human vision works. I expected the "logarithmic" nature of the curve would naturally place middle perceived grey into the middle of the curve!

What should certainly stick is that your card should be rendered of the same grey as it is. Or do you mean it is not?

 

[Mr Bill]

Can somebody(ies) sum up in relatively simple terms that the bulk of APUGers will understand, what RobC has said that will mislead a user of his article to sufficiently wrongly expose a negative? Isn't this the only crucial way to look at what RobC is trying to do for those who wish to get the right or as right as makes no difference the exposure?

I thought I was fairly clear in my early posts (#2, #5, #10), but perhaps I was being too polite about it. The whole thing that Rob has come up with about the "K factor" implying (or proving?) that a reflective meter is calibrated for 8% reflectance is complete fiction. Since this is the foundation of his explanation, everything built on top will also collapse to the same extent.

Rob initially demeaned my objection as being based on an "old" standard (ANSI, too) that doesn't matter any more. But I pointed out, perhaps not bluntly enough, that the current ISO standard he presumably is using is also old - from 1974.

After that, he argued that he is not interested in the standards, only practical results (or something to that general effect). Yet he still continued to base his conclusion on some integral part of the standard(s), the meter formula which includes division by the K constant. In other words, he can use excerpts from the standards to invent his argument, but no one else is allowed to use the standards to argue against him.

In truth, whether you meter from an 18% Kodak card or from an 8% reflectance card, you're probably in the ballpark for an okay exposure. So if your goal is simply to get a reasonable exposure, and you don't care if the underlying reasoning is correct or not then I guess you can just take his article and go with it. But I hope you don't pass the "reasoning" on to other photographers, who may be more finicky about what they want to believe.

If you want to dig deeper into the truth behind this, Stephen has attached the actual ANSI standard. If you go back into the appendices you will find the derivation of the K constant, and if you work at it hard enough, you should reach the understanding that scene reflectance is not involved at all. In fact, scene reflectance is not part of this standard at all. HOWEVER, you might realize that since the standard covers both incident AND reflective meters, there's an implied relationship (not strictly part of the standard) as a result of comparing the K constant (reflective meter) to the C constant (incident meter).

Ps, if this is not clear enough, I'd like to turn your question back around on you. My friend just found 15 cents in his pocket. He says says that this is 15% of a dollar, and coincidentally is about what he thinks an exposure meter is calibrated to - 15% scene reflectance. Is this coincidence? He doesn't think so; the numbers are staring us in the face. He says that THIS is where the meter calibration comes from, and by way of proof, he challenges you to PROVE HIM WRONG! This is the situation I want to turn your original questions toward. Do you still think it is only enough that you get a good exposure from his method? Or do you think the burden should be on my friend to support his contention? (I think the burden is on the guy to support his seemingly preposterous idea.) The two situations are not that different.

 

[Stephen Benskin]

I thought I was fairly clear in my early posts (#2, #5, #10), but perhaps I was being too polite about it. The whole thing that Rob has come up with about the "K factor" implying (or proving?) that a reflective meter is calibrated for 8% reflectance is complete fiction. Since this is the foundation of his explanation, everything built on top will also collapse to the same extent.

Rob initially demeaned my objection as being based on an "old" standard (ANSI, too) that doesn't matter any more. But I pointed out, perhaps not bluntly enough, that the current ISO standard he presumably is using is also old - from 1974.

After that, he argued that he is not interested in the standards, only practical results (or something to that general effect). Yet he still continued to base his conclusion on some integral part of the standard(s), the meter formula which includes division by the K constant. In other words, he can use excerpts from the standards to invent his argument, but no one else is allowed to use the standards to argue against him.

In truth, whether you meter from an 18% Kodak card or from an 8% reflectance card, you're probably in the ballpark for an okay exposure. So if your goal is simply to get a reasonable exposure, and you don't care if the underlying reasoning is correct or not then I guess you can just take his article and go with it. But I hope you don't pass the "reasoning" on to other photographers, who may be more finicky about what they want to believe.

If you want to dig deeper into the truth behind this, Stephen has attached the actual ANSI standard. If you go back into the appendices you will find the derivation of the K constant, and if you work at it hard enough, you should reach the understanding that scene reflectance is not involved at all. In fact, scene reflectance is not part of this standard at all. HOWEVER, you might realize that since the standard covers both incident AND reflective meters, there's an implied relationship (not strictly part of the standard) as a result of comparing the K constant (reflective meter) to the C constant (incident meter).

Ps, if this is not clear enough, I'd like to turn your question back around on you. My friend just found 15 cents in his pocket. He says says that this is 15% of a dollar, and coincidentally is about what he thinks an exposure meter is calibrated to - 15% scene reflectance. Is this coincidence? He doesn't think so; the numbers are staring us in the face. He says that THIS is where the meter calibration comes from, and by way of proof, he challenges you to PROVE HIM WRONG! This is the situation I want to turn your original questions toward. Do you still think it is only enough that you get a good exposure from his method? Or do you think the burden should be on my friend to support his contention? (I think the burden is on the guy to support his seemingly preposterous idea.) The two situations are not that different.

Nicely summed up.

I'm putting together something that might help with understanding K.

 

[RobC]

Diapositivo
I have struggled with this very thought for a long time. But I have not had the courage as you have to express myself for fear of getting a ton of coal or snow (as may be from either end of the scale) poured on me. Having you express it gives me courage.
Key, I think, is " taking fixed the upper value" which is of course never fixed. 100% luminance in Santorini at noon is very different from 100% luminance in my home town on the foggy day yesterday.
Perhaps the wording "18% grey card" or "middle grey" is inappropriate for it's use; I think of that as just a Proper name. Another name might be "this is a tone of card that I can use to represent a tone in my finished print".
I thought of a mind game last night while I was unable to sleep pondering this. I cut a long 21 step Stouffer Step Wedge in thirds taped them all up on a window with clear north sky behind and photographed each one. middle grey would surely be different for each but the exposure must surely be the same. (? presumably based on a spot reading of step 10 of 11)

Anyway, like Diapositivo, this stage of the argument is where I get stuck. (which is like the first paragraph of the argument, so I don't get far)

You should think of a kodak grey card as a reference which is 2 1/2 stops less reflectance than 100% reflectance. Your meter doesn't know what colour it is. It could be any colour. The point is that it reflects 18% of the light falling on it. BUT there is a big caveat to that which is it must be set at the correct angles otherwise its completely useless as a known reference. And as far as I can tell, virtually no one uses it at the correct angles so it is in fact a complete waste of everyones time and will never give you the accuracy you think it will.

 

[markbarendt]

You should think of a kodak grey card as a reference which is 2 1/2 stops less reflectance than 100% reflectance. Your meter doesn't know what colour it is. It could be any colour. The point is that it reflects 18% of the light falling on it. BUT there is a big caveat to that which is it must be set at the correct angles otherwise its completely useless as a known reference. And as far as I can tell, virtually no one uses it at the correct angles so it is in fact a complete waste of everyones time and will never give you the accuracy you think it will.

Rob you ID some real challenges.

There are very practical uses of the card though. For a given individual, the exact angle doesn't matter as long as it's used consistently and the user gets a reliable result. "Known" targets are great tools.

 

[BrianShaw]

You should think of a kodak grey card as a reference which is 2 1/2 stops less reflectance than 100% reflectance. Your meter doesn't know what colour it is. It could be any colour. The point is that it reflects 18% of the light falling on it. BUT there is a big caveat to that which is it must be set at the correct angles otherwise its completely useless as a known reference. And as far as I can tell, virtually no one uses it at the correct angles so it is in fact a complete waste of everyones time and will never give you the accuracy you think it will.

... and Kodak in Professional Photoguide and Roger Hicks in Perfect Exposure agrees. That is in the Kodak instructions for the gray card and the Kodak Light-Ratio Calculator. Seems to me that this is a procedural error based on failure to read instructions.

I don't know how many people know the right procedure or not, but I noticed even some respectable universities publishing instructions to their students indicating that the card should be parallel with the camera lens (and assuming the metering was being done with an in-camera meter).

Maybe what's needed is a report that re-iterates the correct procedure and demonstrates the amount of error when there are deviations from the proper procedure.

EDIT: Just made a few quick measurements of gray card at two angles. Used the gray card in a Kodak Professional Photo Guide (Can't seem to find my full-sized gray card... I never found it very useful but I know I didn't throw it away) using a Gossen L-558 spotmeter and indoor lighting from a north-facing window. Sky has high stratus cloud so I tried to work fast; 6 readings per condition. With the card facing the meter at a perpendicular angle the average reading was 7.2 EV(100) and at a 45 degree angle the average reading was 7.6 EV(100)... so the average error due to gray card angle seems to be about .5 stop.

The average reading measured as incident was 7.1EV

 

[Stephen Benskin]

K is a calibration constant and is composed of three basic parts: light loss in an optical system, characteristics of the exposure meter particularly the spectral characteristic of the photo cell, and a basic constant that is determinant in exposure placement. The meter is calibrated to a known Luminance value and is connected to the film exposure through film speed and K. In use, differences in the target Luminance are compensated with the exposure calculator and the reciprocal nature of the shutter and f-Stop relationship.

According to section 4.1.1 in the ANSI standard, “The instrument-computer combination shall be calibrated in accordance with the formulas in Table 1.

The exposure constants K and C may be assigned values in accordance with Table 1 unless the design parameters differ from the assumptions explained in Appendix C. In this case, new values may be computed as outlined…The manufacture may choose the value which results in optimum exposure for the most critical film, usually color film.”

Please note that some of the values in the example are slightly off due to rounding.

Table 1 basically uses eq. 1 in the below example for reflected exposure meters. To find the value of L, the equation can be reduced down by using Sunny 16 where the shutter speed is the reciprocal of the film speed (eq. 2). 16^2 = 256 and K = 12.5, the value for L = 3200 cd/m^2 (eq. 3).

Now plug 3200 into the camera exposure equation excluding shutter speed and the general exposure value for Eg, eq. 4, is around 8.12. Include the shutter speed for keeping the reciprocal of a 125 speed film, and the resulting exposure for Hg = 0.065 lxs (eq. 5). The final three equations define what Hg should be for each film speed. Eq. 6 uses a target at 800 cd/m^2. By opening up the aperture by two stops, Hg remains at 0.065 lxs.

No matter what the exposure meter is aimed at, it will want to make an exposure Hg = 8/ISO speed.

Here is an excerpt from Appendix C1 in the ANSI standard for proof.

Appendix C1 Exposure Parameters (excerpt)

In order for a meter to be used to set a camera to the proper exposure, the following relationship is assumed to exist:

Hg / Hm = constant (for B&W 125 speed: 0.064 / 0.0064 = 10)

(for color reversal, was 0.064 / 0.064 = 1, current 0.064 / 0.080 = 0.8)

To write this relationship in a usable form to obtain the film exposure in the camera in terms of American National Standard film speed. Thus,

Hg = K1 / Sx (100 8/100 = 0.080 125 8/125 = 0.064 400 8/400 = 0.020)

(K1 = 8), (K1 is what Connelly denotes as P) (K`1 is the variable in the K equation)

This factor K1 has been determined experimentally by psychometrically selecting the “preferred exposure” for scene types, light levels, and camera and meter types covering the ranges normally encountered. It’s value for the purpose of designing a specific exposure control is dependent upon three variables:

1. The spectral characteristics of the photodetector
2. The photographic effectiveness of the scene illuminator
3. The distribution of luminance levels in the scene as measured by the detector

 

[Stephen Benskin]

... and Kodak in Professional Photoguide and Roger Hicks in Perfect Exposure agrees. That is in the Kodak instructions for the gray card and the Kodak Light-Ratio Calculator. Seems to me that this is a procedural error based on failure to read instructions.

I don't know how many people know the right procedure or not, but I noticed even some respectable universities publishing instructions to their students indicating that the card should be parallel with the camera lens (and assuming the metering was being done with an in-camera meter).

Maybe what's needed is a report that re-iterates the correct procedure and demonstrates the amount of error when there are deviations from the proper procedure.

EDIT: Just made a few quick measurements of gray card at two angles. Used the gray card in a Kodak Professional Photo Guide (Can't seem to find my full-sized gray card... I never found it very useful but I know I didn't throw it away) using a Gossen L-558 spotmeter and indoor lighting from a north-facing window. Sky has high stratus cloud so I tried to work fast; 6 readings per condition. With the card facing the meter at a perpendicular angle the average reading was 7.2 EV(100) and at a 45 degree angle the average reading was 7.6 EV(100)... so the average error due to gray card angle seems to be about .5 stop.

The average reading measured as incident was 7.1EV

 

[BrianShaw]

Thanks, but I'm trying to stay away from egg-head data and stick with practical implications.

So it's half a stop, give or take, to me.

 

[Diapositivo]

EDIT: Just made a few quick measurements of gray card at two angles. Used the gray card in a Kodak Professional Photo Guide (Can't seem to find my full-sized gray card... I never found it very useful but I know I didn't throw it away) using a Gossen L-558 spotmeter and indoor lighting from a north-facing window. Sky has high stratus cloud so I tried to work fast; 6 readings per condition. With the card facing the meter at a perpendicular angle the average reading was 7.2 EV(100) and at a 45 degree angle the average reading was 7.6 EV(100)... so the average error due to gray card angle seems to be about .5 stop.

The average reading measured as incident was 7.1EV

That's very interesting.
Presumably the light source is at an angle with the subject-camera axis.
With the grey card perpendicular to subject-camera axis, the result obtained coincides with the incident reading.

The incident reading was obtained using the disc or the sphere?

When working with slides 0.5 EV is a sizeable difference, it makes a big difference in the picture and determines the success or the failure of the shots with some subjects.

 

[Diapositivo]

Diapositivo!
I found that if you point the spot meter at something (neutral color) and use that reading for the camera setting for most slide film it would give you a 1.00 D on the film. This can be done by looking at the characteristic curves published by film manufacturers.

OK Now I finally see what you mean.
I was considering the "straight" part of the curve as a different segment.
I found this source:
Dead Link Removed

It shows how "middle grey" is at density 1 if we "cut" the linear part the way he does (using a part of the toe as well).

The exposure value is also "0" in both cases. So the center of the graphs should be 0, 1.0 in both cases.

Is there a specific reason for it, or is it just a coincidence?

The ISO-ANSI standard, if I get it right, calculates sensitivity by measuring a certain given value of density above base+fog.

 

[BrianShaw]

Sphere. The light axis (a large window) was behind my imaginary camera position.

That's very interesting.
Presumably the light source is at an angle with the subject-camera axis.
With the grey card perpendicular to subject-camera axis, the result obtained coincides with the incident reading.

The incident reading was obtained using the disc or the sphere?

When working with slides 0.5 EV is a sizeable difference, it makes a big difference in the picture and determines the success or the failure of the shots with some subjects.

 

[Sirius Glass]

Thanks, but I'm trying to stay away from egg-head data and stick with practical implications.

So it's half a stop, give or take, to me.

Close enough for photographic work.