The mystery of different values of C in incident light exposure meters

[Doug Kerr]

I return to the original topic of this thread, which was "Why do incident light exposure meters offering both cosine (disk) and cardiod (dome) configurations typically have different values of "C" for the two configurations?"

Of some interest is this passage from Don Norwood's 1961 patent (US 1,183,986) (on an improved light collector for exposure meters):

At the time that the device disclosed in my prior Patent No. 2,214,283 [1940‑09‑10] was designed, it was considered desirable by most photographers that the camera lens settings used under front and side lighting conditions differ from one another by one full f-stop, assuming that the intensity of the light source remained the same.

I am not sure in what context that recommendation was offered. But it almost certainly pertained to the lighting of human subjects (perhaps most notably human heads).

But Norwood goes on to say:

Since the issuance of that patent, however, the characteristics of films have changed considerably, and the opinions of photographers as to what lighting effects are aesthetically most desirable have also changed. With particular reference to the relationship between front, side and back lighting, the film manufacturing companies are now suggesting a differential of only one-half f‑stop in camera setting as between front light and side lighting, and between side lighting and back lighting.

Many expert photographers feel that the proper differential is two-thirds of an f-stop between these various steps, but at any rate it is clear that the full f-stop differential previously recommended should no longer be used.

The hemispherical collector used in an incident light exposure meter following Norwood's concept would have a theoretical directivity of exactly 0.5 at 90°, so with such a meter "aimed" at the camera, it would (as predicted by the "older" rule of thumb) properly reckon the photographic exposure that should be used for a shot lit from the side (assuming that this meter gives an "appropriate" photographic exposure recommendation for light from the camera location).

Now, as to the Sekonic L-398A espsure meter, the manual advocates the use of the "disk" (cosine) configuration for the photography of manuscripts, paintings, and similar "planar" subjects. But we have no idea what Sekonic's "exposure objective" was for this class of work.

Now if we assume the same "exposure objective" for shooting a human face, the we would expect the "C" for the dome configuration to be the same "C chosen for the disk configuration, but it is not - it is about 0.4 stops "hotter".

This might come from the fact that a photo of a side-lit human face is quite different than a photo of a painting, and so the subjective concept of "properly exposed" might jut call for different values of "C".

Recall that this is part science, part art, part craft, and part a certain amount of good luck.

Remember, the overriding rule is, "If the photographic exposure determined this way results in an image that is 'too light' or "too dark', shoot it again with a lesser or greater exposure.

Best regards,

Doug

 

[Doug Kerr]

I note that the manuals for a number of incident light photographic exposure meters (which have dome - cardioid - an disk - cosine light collector advocate the use of the dome collector for most photographic use but the disk collector for photographing for example, manuscripts, paintings, and other 'flt" subjects.

It would seem that the rationale for this is as follows. If we assume that the 'flat" object is a Lambertian reflector, then the luminance of any element is proportional to the illuminance upon the object times the reflectance of that element.

So we are interested in the illuminance upon the object. Our meter, with the disk collector in place (giving it a "cosine" directivity) will determine that, even if the illumination comes from various different directions.

What if we used the meter with the dome collector in place? Its directivity pattern is nominally a cardiod pattern, not cosine. Thus, we cannot expect it to respond to the actual illuminance upon our subject if the illumination comes from other than the camera position.

Now suppose that, in a certain case, we have a flat object illuminated by a single source at the camera position. We might think that in this case, the exposure meter would give the same result (photographic exposure recommendation) with either type of collector aboard.

But in fact with such a meter as the Sekonic L-398A, we cannot expect that, as the manufacturer states a significantly-different value of C (the incident light expsure metering calibraion constant) with the two different collectors in place.

This is thus an illustration of the curiosity of which I speak.

Best regards,

Doug

 

[Doug Kerr]

I have today been perusing Exposure Manual, Third edition (1974), by J. F. Dunn and G. L. Wakefield. This is a large book (about 250 pages) and is very admirable. The descriptions are technically-thorough, but do not involve esoteric concepts or mathematics. I will speak below if Dunn were the sole author; the passages I cite are from his portion of the book.

Dunn speaks at some length about the "duplex" exposure metering technique. He describes it as being done with the geometric average of the two luminance measurements.

He then discussed what we may call the "Norwood" exposure metering technique. He tells that extensive tests were done with a wide variety of scenes, comparing the "duplex" and "Norwood" photographic exposure recommendations. He tells that for the most part those agreed within 1/3 stop.

He tells that in this, the calibration constant, C, was essentially the same for the "cosine" meter used for the duplex measurements and the "Norwood" (cardioid) meter.

Best regards,

Doug

 

[Doug Kerr]

J. F. Dunn tells us, in Exposure Manual, that in extensive tests over a wide range of scenes, the photographic exposure recommendation given by the "Norwood" exposure metering system and that given by the "duplex" metering method generally agreed within 1/3 stop. I note that seemingly, the "duplex" method employed by Dunn used the geometric average of the two luminance (or quasi-luminance) measurements.

I of course have no way to duplicate that series of tests, nor does that scenario lead easily, if at all, to any theoretical model of the lighting situations.

But what I can do is model the expected results of the two metering systems in a series of cases that Don Norwood initially emphasized in his work: the illumination of a subject with the "key-fill" lighting technique.

I have simulated, in an MS Excel spreadsheet, the expected results of both metering techniques for a series of cases involving (a) various positions of the key light and (b) various key"fill ratios.

The table below is the summary of the results of those simulated tests:

The values in the table are the ratio of the photographic exposure recommendation of the Norwood system measurement to that given by the "duplex" measurement. They are here based on the two meters having identical values of "C", the incident light exposure metering calibration constant.

We see that, as the position of the key light reaches 90°, for any key:fill ratio the two measurement results differ by more than 1/3 stop. But for lesser key light angles (such as 60°), the two measurement results differ by less than 1/3 stop.

I do not mean here to imply that the one or the other of these measurement technique gives "better" results. But seemingly, before the onset of the "Norwood" system, cinematographers used the "duplex" metering system with satisfactory results, and then found that the easier to use "Norwood" system gave equally-satisfactory results.

But it is likely that the scope of that comparison was over subjects lit by the key-fill lighting technique.
******
It is interesting that when the same simulation was done based on using the arithmetic average of the two meter readings in the duplex method, the extent of agreement between the "duplex" method and the "Norwood" method was what we might conmsider to be overall better, as we see from this table:

Best regards,

Doug