Duplex metering according to J. F. Dunn

[Doug Kerr]

Noted photographic exposure and exposure metering authority J. F. Dunn, in a 1948 paper before a major British cinematography society, begins by laying out some of the objectives of exposure metering and "proper" exposure control. He points out some of the practical difficulties in playing this game.

He then introduces the concept of "duplex metering", which he says can deal conveniently with those difficulties.

In any case, his context was not the specialized context of Norwood's original work (the use of the key-fill lighting technique) but rather a more general photographic context. His examples are largely drawn from the situation of photography by "daylight". The technique as he describes it uses an exposure meter with a flat (cosine directivity) receptor, which I note ideally is responsive to the true illuminance of the overall illumination on the plane of the meter's receptor.

He then says that two readings should be made with this meter, located at the subject, the first with the meter's receptor facing the primary source of ;light (the sun, in the "daylight" case) and the second with the receptor facing the camera. The geometric average of these two readings is reckoned and set into the meter's calculator as if it were the meter reading. The photographic exposure recommendation then given by the calculator would be used on the camera for the shot.

Except that, very likely, says Dunn, one would have to use one stop less photographic exposure to get the proper exposure result.

******
But then, in his later book, Exposure Manual, Dunn tells us that an even handier way to determine the proper photographic exposure for a scene was the use of a "Norwood system" exposure meter, with a hemispherical light collector (or equivalent). This is used to make a single measurement (with the dome of the meter facing the camera, and you are good to go.

Dunn then tells that in a series of experiments, the photographic exposure recommended with the "Norwood system" meter was within 1/3 stop of that given with the duplex metering technique. (He does not mention whether the duplex metering used a one stop "bump" of not.)

Very interesting.

Best regards,

Doug

 

[Doug Kerr]

Something I missed in Dunn's original description of the duplex metering system, in the outdoor context, is that the "sun" measurement was not to be taken with the meter aimed at the sun (as I had described) but rather in a horizontal plane but at the azimuth of the sun.

Exactly how that works I have no idea.

In his later book, Exposure Manual, when Dunn describes the duplex metering system he says that, in the outdoor case, one measurement is to be made "in the direction of the sun". One might of course assume that this means "toward the sun", but it might also be that he meant (without making it clear) "at the azimuth of the sun (but in the horizontal plane)".

Best regards,

Doug

 

[Doug Kerr]

Some further insight can be gotten from this entire page from Dunn's book (used here under the doctrine of fair use). Here, images 1 and 2 were apparently taken with the photographic exposure being that coming directly from that meter measurement, while image 3 was taken with the photographic exposure implied by the geometric mean of the illuminances given by the first two readings.

This seems to be for a case where the sun is at an angle of 135° to the camera-subject and the "sun" reading is made with the meter aimed at the sun itself. (We see the person working the meter in shots 1 and 2.)

Certainly "in the usual sense" the third image is the most "pleasing".

Much fun.

Best regards,

Doug

 

[Doug Kerr]

I continue to be somewhat mystified by the entire matter of "duplex metering".

For example, in the second shot in the page shown above, I would expect that it we took a basic incident light reading, using a meter calibration with a value of C that we ordinarily find suitable, the male subject would have been "properly" exposed.

The "brightness" of any element of the subject is dependent on its reflectance and on the illuminance upon it (however the latter comes about). So if the meter measures that illuminance, the exposure should come out "well".

Yet here we see him seemingly significantly "overexposed".

Now I call attention to the note that suggests that when this metering technique is being used, we probably should set the film sensitivity into the exposure calculator at one stop less that its rated value. The effect of this is essentially to make the meter read one stop "too hot" in its photographic exposure recommendation. I suspect this was done for these example shots. And the result is that in the second shot, the subject is overexposed.

But here we are told that the way to avoid this overexposure is to use duplex metering, by way of which the illuminance value set into the meter's calculator is (in this example) is about 1.7 stop greater than if we had used the simple incident light measurement (with the meter facing the camera). (This led to a photographic exposure recommendation of 1/50 sec @ f/4 rather than 1/50 sec @ f/2.2.)

Teiku.

Best regards,

Doug

 

[Doug Kerr]

The examples do not speak of the actual illuminance readings of the meter for the two duplex measurements, but rather of the photographic exposure recommendations given by the two readings (the film sensitivity presumably having been set consistently).

And since the photographic exposure recommendations are all stated in terms of the same shutter speed (1/50 sec), we can conclude the measured illuminance values from the apertures.

Doing that and using the "geometric mean" calculation, with the two "input" values being f/7 and f/2.2, the resulting aperture result would calculate as about f/3.9 (which is appropriately rounded to f/4).

How did I do that? Well, the exposure effect of an aperture of a certain f-number is inversely proportional to the square of the f-number. So I square 7 and get 49, and I square 2.2 and get 4.84.

To get the geometric mean of those numbers I multiply them, giving 237.16, and take the square root of that, giving 15.4. But because we want our result to be in terms of the resulting recommended aperture, we take the square root of that, giving 3.92 (f/3.92).

Best regards,

Doug

 

[Ted Cousins]

... since the photographic exposure recommendations are all stated in terms of the same shutter speed (1/50 sec), we can conclude the measured illuminance values from the apertures.

Doing that and using the "geometric mean" calculation, with the two "input" values being f/7 and f/2.2, the resulting aperture result would calculate as about f/3.9 (which is appropriately rounded to f/4).

How did I do that? Well, the exposure effect of an aperture of a certain f-number is inversely proportional to the square of the f-number. So I square 7 and get 49, and I square 2.2 and get 4.84.

To get the geometric mean of those numbers I multiply them, giving 237.16, and take the square root of that, giving 15.4. But because we want our result to be in terms of the resulting recommended aperture, we take the square root of that, giving 3.92 (f/3.92).

Made me wonder about EV, Doug, seein' as most meters have an EV scale.

f/7 at 1/50 sec: log2(49x50) --> 11.29 EV

f/2.2 at 1/50 sec: log2(4.84x50) --> 7.92 EV

How to get from there to f/3.92, I wonder?

looks like the arithmetic mean of the EVs does it:

log2(3.92x50) --> 9.59 EV, close enough for photographic work ...

... and no need for the same shutter speed??

In other words:

• point the meter in direction a) note the recommended EVa
• point the meter in direction b) note the recommended EVb
• figure the arithmetic mean EVm and set the calculator EV to that
• Choose the desired aperture/shutter combination

best, Ted.

 

[Doug Kerr]

Hi, Ted,

Made me wonder about EV, Doug.

f/7 at 1/50 sec: log2(49x50) --> 11.29 EV

f/2.2 at 1/50 sec: log2(4.84x50) --> 7.92 EV

How to get from there to f/3.92, I wonder?

best, Ted.

Well,1/50 sec at f/3.92 is log2(15.36x50) = Ev 9.58.

The geometric mean of two photographic exposures corresponds to the arithmetic mean of their Ev. So if our two measurements indicate Ev 11.29 and Ev 7.92, the geometric mean of their exposures would be about Ev 9.6.

Best regards,

Doug

 

[Ted Cousins]

Hi, Ted,

Well,1/50 sec at f/3.92 is log2(15.36x50) = Ev 9.58.

The geometric mean of two photographic exposures corresponds to the arithmetic mean of their Ev. So if our two measurements indicate Ev 11.29 and Ev 7.92, the geometric mean of their exposures would be about Ev 9.6.

Best regards,

Doug

As just now edited, my point being that the complication of restricting the calc. to the same shutter speed becomes unnecessary.

best,

Ted.

 

[Doug Kerr]

Hi, Ted,

So if we wanted to do duplex metering, using the "geometric mean" algorithm, and wanted to work in Ev, we would:

• Take the two measurements and note the photographic exposure recommended from each in Ev.

• Take the arithmetic mean of the two Ev values.

• Use the resulting Ev as the camera settings.

Best regards,

Doug

 

[Doug Kerr]

Hi, Ted,

As just now edited, my point being that the complication of restricting the calc. to the same shutter speed becomes unnecessary.

Sure.

Working on an Ev basis makes the whole process simpler.

Best regards,

Doug

 

[Ted Cousins]

As just now edited, my point being that the complication of restricting the calc. to the same shutter speed becomes unnecessary.

Hi, Ted,

So if we wanted to do duplex metering, using the "geometric mean" algorithm, and wanted to work in Ev, we would:

• Take the two measurements and note the photographic exposure recommended from each in Ev.

• Take the arithmetic mean of the two Ev values.

• Use the resulting Ev as the camera settings.

Best regards,

Doug

Did you miss my earlier edit, Doug?

In other words:

point the meter in direction a) note the recommended EVa
point the meter in direction b) note the recommended EVb
figure the arithmetic mean EVm and set the calculator EV to that
Choose the desired aperture/shutter combination

Well, at least we said the same thing ...