Norwood's dome lives on

[Doug Kerr]

Donald Norwood's initial quest was for a system that would facilitate the determination of the "proper" photographic exposure during the photography of a human face (head, actually) using the "key-fill" lighting technique, ideally allowing this to be done with a single measurement with an appropriate "incident light" instrument.

In support of his approach to this, he ran an elaborate series of tests that showed, as the key light (with a certain fixed "potency" at the subject) was successively moved to positions offset from the camera by various angles, in order to attain what was judged by observers to be a "consistent" exposure result, the photographic exposure (combination of shutter speed and aperture) had to be increased by a certain amount depending on the angle of the light.

He then converted that to a matter of the "relative effectivity" of the key light (in contributing to "exposure") at different angles. He said that his data well fit a straight-line function (which platted as a directivity curve in polar coordinates gives an "Archimedean spiral"). But it didn't quite; Norwood just plugged that in transcribing one form of the data to another.

Norwood then wen on to tell us that, happily, the theoretical directivity curve of a hemispherical light collector was just that function (it isn't).

But the numerical errors there are not great, and this whole matter of exposure metering is a bit loosey-goosey at best. So incident light exposure meters with hemispherical collectors were found to do that job very well among cinematiographers (and shitly, by potrait phtograhers as well).

But what about other genres of phtography, typical landscape scenes for example.?

Norwoood tells us an outdoor scene can be considered as illuminated by two sources: a primary source, which is the sun, and whose light comes to us from essentially one direction, and a secondary source, the light sky in general, whose light comes to us from every direction over perhaps a hemisphere.

He then goes on to suggest that the light arriving from different directions will be appropriately "discounted" as to its effectivity on exposure by the directivity of the meter's hemispherical collector, based on the angle from which it came.

And, consistent with that, today we find the use of hemispherical-collector incodent light meters highly spoken of for general phtographic use.

Best regards,

Doug

 

[Doug Kerr]

It is perhaps interesting to note that Norwood suggested, for the "baseline" exposure (with the key light at the camera position) values of the incident light metering calibration constant (we denote that as "C"; Norwood called in "K") of from 280-323 (to state it in ISO units, not the units used in Norwood's paper).

It seems common today for incident light exposure meters using a hemispherical ("dome") collector to have values of C of perhaps 330-340. I consider these to be "comparable" to the values advocated by Norwood, especially if we take into account what lens transmission factors may have been in effect in Norwood's camera and "allowed for" in today's exposure meters.

Best regards,

Doug

 

[Doug Kerr]

In most discussions of the underpinnings of Norwood exposure metering scheme, the key light is assumed to be at the same height as the subject (although this is rarely mentioned).

Still, in a "Norwood system" exposure meter, the meter's directivity pattern is rotationally-symmetrical. That is:

• a light source in the same horizontal plane as the meter, located at an angle of 37 degrees from the meter-camera axis, and

• a light source directly in front of the camera, except located 37 degrees above the meter-camera axis

will be treated the same by the "Norwood system" exposure meter (we assume the meter to be "pointed at" the camera). What matters to the meter is the angle between:

• A line from the meter to the light source, and

• A line from the meter to the camera (we again assume the meter to be "pointed at" the camera)

If we should feel the need to look into this mathematically, that angle can be readily computed. We start with the more general formula derived from the spherical law of cosines, which is a bit complicated. But in this case, the situation is such that one term becomes zero and vanishes, and one factor of the other term becomes 1 (and thus disappears from the notation).

The resulting formula is:

A=arccos(cos(H)cos(V))

where A is the angle of the light source (with respect to the meter-camera axis). H is the horizontal angle of the light source (we might say its "azimuth"), with respect to the camera, and V is its vertical angle (we might say its "elevation").

Best regards,

Doug

 

[Doug Kerr]

The above suggests that, in two shots of a human head, with:

• the key light at the same height as the subject/meter, but directly to one side (A=90°), and

• the key light directly over the subject/meter (again A=90°)

the "Norwood system" exposure meter will give the same photographic exposure recommendation, and we are to believe that the resulting two shots will appear to be "equally exposed".

Best regards,

Doug

 

[Doug Kerr]

A major underpinning of Donald Norwood's exposure metering system is a series of tests of head shots of several human models, lit by the "key-fill" technique.

For each model, a series of shots were taken, first one with the key light "on camera" (and the exposure st to what was considered to produce a "good", result, and then with the key light moved to various locations (to the side). For each of these positions of the key light, a number of shots ere taken, with photographic exposures in half-stop steps (I think both greater than and less than the initial "reference" shot).

The negatives were all developed in a consistent way and prints made with a consistent enlarger exposure (I think).

A panel of viewers then were asked, for each subject, for each (to the side) key light position, which of the shots looked "equally exposed" compared to the reference shot (with the key light "head on") (and thus would be considered "properly exposed" as well).

Norwood assures us in his paper describing this test program that:

"It was not difficult to note which pictures appeared 'right,”'and matched in appearance."

Certainly the overal appearnce of ths subject's face would be signficantly different in the shots with the key light in varous locations, as a resuilt of the chage in "sculpting" that the different key light locarions produced.

I have not had the privilege of examining images from Norwood's test, but it seems to me that it would be very difficult to consider two prints of shots taken with the key light at different locations and judge which ones were "equally exposed".

Just sayin'.

Best regards,

Doug

 

[Doug Kerr]

By way of explanation, I will show here four shots of a model's face taken from Art Adams' nice online essay on key-fill lighting. They are in order left-to right, then down.

The first shot was with the key light "head on". In the following shots, the key light (I suspect of consistent
"potency"') was moved to successive angles to our right side, with the last shot probably with the key light at 45°.

There is no discussion of exposure control ion the article, sop I have no idea how the photographic exposure may have varied between these shots.

Now here is the issue: If I were asked to compare shots 1 and 4 and say whether I considered them having "equal exposure result), or to compare them as to the degree that the exposure result was "good", I'm not sure how I would do that.

Best regards,

Doug

 

[Ted Cousins]

View attachment 14477 View attachment 14483

The first shot was with the key light "head on" with the last shot probably with the key light at 45°.

There is no discussion of exposure control ion the article, sop I have no idea how the photographic exposure may have varied between these shots.

Now here is the issue: If I were asked to compare shots 1 and 4 and say whether I considered them having "equal exposure result), or to compare them as to the degree that the exposure result was "good", I'm not sure how I would do that.

One way, Doug:

I observed linear histograms in the GIMP ... same mean value 0.422 and median 0.314 - but "head on" is std. dev. 0.220 vs. 0.208 at "side". Significantly more contrast "head on".

So the exposure results were about the same and I preferred the side-lit look.

best,

Ted.

 

[Doug Kerr]

Hi, Ted,

One way, Doug:

I observed linear histograms in the GIMP ... same mean value 0.422 and median 0.314 - but "head on" is std. dev. 0.220 vs. 0.208 at "side". Significantly more contrast "head on".

Thanks for that. Very interesting.

So the exposure results were about the same and I preferred the side-lit look.

Which one in particular?

Assuming I am not going for some special artistic effect, I think I would prefer number 2.

Best regards,

Doug

 

[Ted Cousins]

Hi, Ted,

Thanks for that. Very interesting.


Which one in particular?

The second attachment in my post ... "A", your "shot 4" ... you had asked about 1 versus 4.

best,

Ted.