Hiding in Plain Sight

Bill Burk · Jan 25, 2012

The way time and speed cancel each other out and aperture becomes a constant, 16 that is easy to square.

Eerily coincidental. Or as you say not a coincidence at all.

Did Sunny 16 work the same before and after the safety factor was removed?

Stephen Benskin · Jan 25, 2012

You need to think about each part on it's own. The scene luminance range and illuminance doesn't change, the exposure meter doesn't change, it's the placement of the exposure on the film curve that changes based on the speed point / exposure meter ratio.

Bill Burk · Jan 25, 2012

Stephen Benskin said:
You need to think about each part on it's own. The scene luminance range and illuminance doesn't change, the exposure meter doesn't change, it's the placement of the exposure on the film curve that changes based on the speed point / exposure meter ratio.

But if it has always been f/16 and shutter = reciprocal of EI... What changed when the safety factor was removed.

Let me think a minute.

It was ASA (old standard with safety factor) 50 so f/16 at 1/50th... Now it's ISO (current standard) 100 so f/16 at 1/100th...

OK the exposure did change.

Stephen Benskin · Jan 26, 2012

Bill, I like to come at the problem from many different directions, and if the answer is still the same, there's a good possiblity it's correct.

Everything in photography is interconnected. A piece of information might seem to work on its own, but plug it into the bigger picture and it doesn't quite fit. Like with the reflected exposure meter reading 18%. That sounds right until you compare it with the incident meter's illuminance. A normal negative should have a density range of 1.25 to fit on a grade 2 paper sound good until you compare that to the paper LER range of 1.05 for a grade 2 paper. On the other hand, I'm happy to say the Sunny 16 seems to be able to hold up under scrutiny.

That's why I really like the constants equation. It shows the interconnection between the vaious elements involved in the photogaphic exposure by combining the constants from the exposure meter, K, the camera, q, and the photogaphic exposure, P, into an equation.

K * q = P or P / q = K or whatever variable you're solving for. K is in cd/m^2 or nits.

12.48 * .65 = 8.11

Bill Burk · Jan 27, 2012

It's in the Journal of Photographic Science, Vol. 16 1968 page 188...

Talking about the arithmetic speed where the first way is sensitometric testing and the 0.1 speed point... "In the second way, Sa is determined by a photographic test, and is equated to the reciprocal of the effective exposure time when a satisfactory photograph is taken with a camera set at f/16, on a clear sunny day with a solar altitude of about 40 degrees."
...
"It is evident that the two methods of film speed assessment determine in effect the ratio..."

"k1 = Hg/HM"

---
So...

It almost seems like you can say "The arithmetic film speeds were setup that way to keep Sunny 16 valid."

Then perhaps indirectly since by this time there is a tie-in to the sensitometrically determined speed point...

(Another thread) "The metering point was chosen to align the average scene to agree with Sunny 16 results"

And what I kind of find funny: The ratio for Sunny 16 seems to solve to K but why? It is an "incident" light scenario - shouldn't it relate to C?

Stephen Benskin · Jan 27, 2012

Bill Burk said:
Talking about the arithmetic speed where the first way is sensitometric testing and the 0.1 speed point... "In the second way, Sa is determined by a photographic test, and is equated to the reciprocal of the effective exposure time when a satisfactory photograph is taken with a camera set at f/16, on a clear sunny day with a solar altitude of about 40 degrees."
...
"It is evident that the two methods of film speed assessment determine in effect the ratio..."

"k1 = Hg/HM"

You could even say that the sensitometric speed point is about the characteristics of the film curve (lower gradient limits with B&W negative) and is represented by Hm, and the photographic testing (think psychophysical testing) sets the value of the speed constant.

So...

It almost seems like you can say "The arithmetic film speeds were setup that way to keep Sunny 16 valid."

Then perhaps indirectly since by this time there is a tie-in to the sensitometrically determined speed point...

(Another thread) "The metering point was chosen to align the average scene to agree with Sunny 16 results"

And what I kind of find funny: The ratio for Sunny 16 seems to solve to K but why? It is an "incident" light scenario - shouldn't it relate to C?

You're presenting some interesting questions here. You should flesh them out.

Stephen Benskin · Jan 30, 2012

Bill Burk said:
And what I kind of find funny: The ratio for Sunny 16 seems to solve to K but why? It is an "incident" light scenario - shouldn't it relate to C?

Bill, I've been hoping you expand on this point because it looks like you see something. I'm not sure exactly what. Sunny 16 is about the the average illuminance which has to do with C, and K has to do with the average luminance which is a product of the average illuminance. They are interconnected.

Bill Burk · Jan 30, 2012

Stephen Benskin said:
Bill, I've been hoping you expand on this point because it looks like you see something. I'm not sure exactly what. Sunny 16 is about the the average illuminance which has to do with C, and K has to do with the average luminance which is a product of the average illuminance. They are interconnected.

OK it makes sense that Sunny 16 should describe average illumination which would be a constant related to the incident meter, and the incident meter constant is C.

And that amount of light falls on an average scene, which has a reflectance that might be 12% or 18% or somewhere in that vicinity. When that average % of light comes back it relates to the constant K.

Is it then oversimplifying to say that from the 0.1 speed point K is the distance to the average reflected scene. And C is the distance to the average sunlight falling upon it?

Stephen Benskin · Jan 31, 2012

Bill Burk said:
Is it then oversimplifying to say that from the 0.1 speed point K is the distance to the average reflected scene. And C is the distance to the average sunlight falling upon it?

The speed point really doesn't factor into the exposure meter constants, and I'm not sure what "distance" means in this context. I can see that you are working out how everything is connected and it all does fit. It will eventually all click together if you keep playing with it.

Both K and C are constants that factor in the elements of the camera's optical system and elements of the exposure meter. K and the exposure equation's constant, q, are closely related and have many of the same variables. The only difference is that q doesn't contain the variables for the exposure meter.

The excerpt below is Appendix C from the 1971 ANSI Exposure Meter standard. You should be familiar with the values and equations.

Bill Burk · Jan 31, 2012

I'm thinking this needs a new graphic diagram.

A diagram could be used to illustrate the 18% and 12% controversy.

Another diagram could show the metered points and K and C. I can't always follow along with math formulae unless they are explained step by step. And I think with diagrams, the concepts could be revealed even more clearly

Stephen Benskin · Feb 1, 2012

Bill, I'm not sure how to do a diagram to illustrate K and C. I might be able to go step-by-step though. As for 12% and 18%, I think I might have a way of looking at it.

The first graph is keyed to 12% reflectance. You can see that the relationship between the speed point and the metered exposure is 10x (1.0 logs). 18% reflectance falls 0.15 log-H further up the curve.

The second graph is keyed to 18% reflectance. The shadow exposure, including flare, now falls 1/2 stop further down on the curve.

Stephen Benskin · Feb 2, 2012

The 18% graph is somewhat of a cheat. It supposes the same luminance distribution as the 12% model. Of course, this would lead to a different film speed / metered exposure ratio between the two. But what if 18% was middle grey and had the same luminance distribution as the 12% model? The 18% curve would look identical to the 12% one above. One big difference. The math doesn't work for an 18% grey model.

Bill Burk · Feb 3, 2012

I have another theory about how the 18% gray mystique came about.

Kodak make the card to use in print press calibration, four color process, to reveal when a press ink balance is correct (neutral gray). If you make printing ink too dark, it will be harder to judge neutral. So a lighter tone was selected for the purpose of showing color balance.

Then when the idea to use a gray card as an average subject "proxy" - like using the palm of your hand - the idea was given out as "use a known shade of gray, like this 18% gray card". Well the mistake that point forward is that the known shade was not adjusted into the exposure calculation to place it at the known place.

Everybody knew using the palm of your hand was brighter, so you always open up. Probably because the first time you forget to open up your picture is far from expected.

But the gray card was pretty close, so the discrepancy didn't cause much grief.

Stephen Benskin · Feb 4, 2012

What you say sounds possible. No one can know for sure where or how it all began. One of the problems with the question of 18% gray is that there multiple ways it is used. Of which two have to do directly with photographic exposure. Is it the average reflectance and are exposure meters calibrated to it?

I believe the use of 18% gray in printing and two dimensional art has it's origins with Munsell.

Munsell's steps are based on perceived equally spaced steps between black and white with the middle step being 18% gray. Kodak's Gray Card Plus has a white and a black area included. According to their information, the white has a 90% reflectance (RD 0.046), and the black has a 3% reflectance (RD 1.52). The mean is (1.52-0.046) / 2 = 0.739 Reflection Density or 18% reflectance. So 18% can easily work as the average for a two dimensional subject.

And metering off the hand? Here's a spectral reflectance graph of Caucasian skin. Looks like a high level of infrared reflectance. I guess metering off the hand is more a rule of thumb. sorry.

C and K weren't always the values they are today either.

Stephen Benskin · Feb 5, 2012

Here are some values of K and C through the years. Because there are numerous factors involved in these values, there cant be a perfect one-to-one comparison.

Joint Army-Navy Specification JAN-M-58
Meters, Photographic Exposure (Reflected-Light, Photoelectric Type)
Sept 15, 1944

K = 1.25

ASA Z38.2.6 1948
General Purpose Photographic Exposure Meters (Photoelectric Type)

K = 1 1.35
C = 15 30

ASA PH2.12-1957
General Purpose Photographic Exposure Meters (Photoelectric Type)

K = 1 1.35
C = 22 + - 5

ASA PH2.12 1961
General Purpose Photographic Exposure Meters (Photoelectric Type)

K = 1.06
C = 20.83

ANSI PH3.49 1971
For General Purpose Photographic Exposure Meters (Photoelectric Type)

K = 1.16
C = 30 +- 5

ISO 2720 1974 (R1994)
ANSI/NAPM IT3.302-1994
Photography - General purpose photographic exposure meters (photoelectric type) - Guide to product specification

K1 = 10.6 13.4 (0.985 1.245)
K2 = 13.3 to 16.9 (1.236 1.571)

Cardioid type receptor (hemisphere)
C1a = 320 to 540 (30 50)
C2b = 400 680 (37 63)

Cosine type receptor (flat disk)
C1b = 240 400 (22.3 37)
C2b = 300 500 (28 45)

Stephen Benskin · Feb 6, 2012

According to the 1962 paper An Interpretation of Current Exposure Meter Technology by Allen Stimson,

"The new value K = 3.333 is used when the meters are calibrated at 4700K. The former value K = 3.6, which has been used when the meters were calibrated at 2700K, is 10% greater because the average cell is about 10% less sensitive at this color temperature. Consequently, no radical change in calibration of reputable American made meters is anticipated. However, all meters which conform will be more uniform."

"The incident-light value Iv involves the constant C which can vary +-24% within the limits of the standard and at the option of the meter designer. Incident light, as defined in the Standard for Photographic Exposure, is measured at the subject position in a plane normal to the direction of the camera. However, illumination for visual purposes is ordinarily measured in a horizontal plane. For these reasons, photographic incident-light data may not agree with those of illuminating engineers. The tolerance on the value of the exposure constant K for reflected light meters is less than that on the value of C for incident light meters, because of the greater variation in the acceptance angles and directions of aiming the incident light receivers."

"The ratio of K to C is the average scene reflectance for which the meter is calibrated. the mean value is now R = 3.333 / 20.83 = 16%. This value of reflectance is indicated when the reflected light meter is aimed from the camera toward the subject, and the incident light meter is aimed from the subject toward the camera. However, it is close to the reflectance of the "gray card" used with reflected light meters. Kodak's "Neutral Gray Card" had a reflectance of 18%. The difference is due to the angle at which the card is held."

Today the "ideal" appears to be 3.64 / 30 = 12%.

Here's the technical data for a Sekonic meter.

K = 12.5/10.76 * pi = K = 3.65
C = 340/10.76 = 31.6

Bill Burk · Feb 6, 2012

I'd seen the comment about the angle a gray card is held. Of course you can move the card around and get different readings. What angle were they suggesting to hold the card?

I would normally use a gray card for flat art copying. In the setup I have two light sources at 45-degrees and the camera positioned so the film was parallel to the art. Thinking about flatwork, gets me to thinking about the difference between flat art and three dimensional subjects.

On average, flat art would have a short brightness range; while three-dimensional subjects could have a wider range to consider for its average because three-dimensional subjects include areas in shade (darker) and areas under more light (specular highlights).

Stephen Benskin · Feb 6, 2012

Bill Burk said:
What angle were they suggesting to hold the card?

I believe the second quote from Interpretation of Current Exposure Meters addresses this. The subject is perpendicular on a plane normal to the direction of the camera with a solar elevation of around 41 degrees.

Notice that the earlier standards don't appear to make a distinction between a cosine disk and a cardioid hemisphere receptor? I find that interesting. The difference in C between the two is around .71 to .74. For the Sekonic meter example, Cardioid 340/10.76 * .74 = 23.38, Cosine 250/10.76 = 23.23.

Bill Burk · Feb 7, 2012

Stephen Benskin said:
... The difference is due to the angle at which the card is held ...

This then, still puzzles me. Everything seems to agree that the card faces the camera, parallel with the plane of the film. So I don't know what difference is due to the angle at which the card is held.

Is the sentence, in context, talking about the difference between how a lighting engineer measures illumination versus the photographer's measurement? Obviously, a lighting engineer can make no assumption about where the subject or viewer is. Lighting engineer is merely lighting a room.

Stephen Benskin · Feb 7, 2012

I think he is saying that lighting engineers measure on a horizontal plane, ie ground, with the meter pointed up. I also think Stimpson is suggesting that it's really hard to hold the gray card perfectly perpendicular to the camera plane. We're only talking about the difference between 18% and 16%. With an 18% gray card, the card only needs to be tilted forward about 20 degrees to produce a 12% reflectance when the solar elevation is at 41 degrees (60 degrees total). For 16%, that would be around 5 degrees or so.

Stephen Benskin · Feb 7, 2012

The technical specifications for a Minolta Meter III

K = 14
K = 14/10.76 = 1.3
K = 1.3*pi = 4.08

C = 330
C = 330/10.76 = 30.6

Hemisphere
14*pi / 330 = 13%

Cosine
14*pi / 250 = 17.6%

Again, the early standards don't distinguish the type of receptor for the value of C. Did they use ranges to incorporate both types instead of separating them? Is it valid to use the K/C equation for a cosine receptor? If it is, it apparently helps to support the idea of 18% reflectance for flat art.

No matter what the exposure meter "sees", it puts it all on the same spot on the film - 8/ISO. I believe one of the reasons why the question of the percentage of what the exposure meter reads has so little written about in the more technical publications is because of the variance of conditions and that most falls within tolerance.

Bill Burk · Feb 7, 2012

Michael R 1974 said:
"Do You Know What You're Getting?: Theoretical and Practical Sensitometry for the Analog Photographer".

This would be great! I enjoy the somewhat syncopated conversations, but hopefully the core concepts are reinforced and could be summarized.

Stephen Benskin · Feb 7, 2012

I did a thing a little over a year ago taking exposure from the average illuminance through to exposure placement. It was in the thread "Is the K factor relevant to me or should I cancel it out?" starting at (there was a url link here which no longer exists).

I did give the idea some thought when I was writing the four quadrant reproduction diagram program, but anything beyond something like the K factor thread sounds like too much work for such a niche subject. I'm not sure how many people are interested in this stuff anymore. Few people participate in these discussions, and half of those seem to make derogatory comments. There's also so much I don't understand. The saying that you may think you understand something until you have to explain it to another person is so true.

I do like debunking myths though.

Stephen Benskin · Feb 8, 2012

There is always going to be detractors. I just wish there was more participation in general. There are a lot of smart people on APUG, so the only conclusion is there is a lack of interest on these topic. If people were interested and didn't understand something, would they be asking questions?

I'm glad at least a few people are interested. Michael, you should jump in more. These discussions can really benefit from someone with your experience.

BTW, I believe the posts in the K factor thread are the only time I've ever seen an attempt at connecting the dots between all the different aspects of exposure. In the way information is generally presented, there is a tendency to break concepts down into smaller parts in order to get a handle on them, but then it's hard to see how everything fits together. That's one of the reasons I did the four quad program. Most people work with the film curve and paper curve as seperate objects, and completely leave out the camera image curve. This makes it hard to understand how all the elements of the photographic process interconnect. Concepts become disjointed like the ZS NDR and the paper LER.

I'd like to think the film speed / metered exposure ratio thread connected a few dots too, but without anyone participating how well were they communicated? Maybe with some more feedback, those concepts can be better and more clearly defined.

Stephen Benskin · Feb 8, 2012

Sorry, I typed that last post at work. That last part should read that "without any participation it's difficult to gauge how well the concepts were communicated."

Hiding in Plain Sight

Privacy & Transparency

Privacy & Transparency