Norwood's dome in incident light exposure meters

[Doug Kerr]

Many incident light exposure meters, both electromechanical and electronic, have a light collector in the form of a translucent hemispherical dome, or something similar.

This construct has its roots in a development by Donald W. Norwood in the late 1930s. The story is a fascinating one, with many twists and turns.

The story, best I can reconstruct it, is told in considerable detail in my article on The Pumpkin, "Norwood’s dome: a revolution in
incident‑light photographic exposure metering".

I have just posted to The Pumpkin a revised issue of that article (as Issue 6). There is essentially no new nor substantially-changed information in this issue. It is primarily intended just to improve the presentation.

The article is indexed here on The Pumpkin:

The Pumpkin

Index page of Doug Kerr's The Pumpkin

dougkerr.net

Best regards,

Doug

 

[Doug Kerr]

The basic concept behind the "dome" light collector on incident light photographic exposure meters is this.

A given "potency' of illumination falling on a subject from some direction other than "from the camera position" is "less effective" in creating a certain exposure result than the same "potency of illumination" falling on the subject in the direction from the camera position.

This cannot be predicted with any theoretical model - do not look for ways to prove this that involve cosines or such. This is an empirical reality. This comes about largely because the human visual assessment of when two exposure results are "comparable" depends in an unpredictable way on how a subject's features may be "shaded" by illumination coming from different directions.

But various researchers - including Don Norwood - have by subjective testing determined typically by what degree a certain potency of illumination affects the assessment of the exposure result as a function of its angle of arrival at the subject.

If we make an incident light exposure meter whose directivity (sensitivity as a function of the angle of arrival of the light) follows that function, then a single measurement made with the meter, regardless of the direction from which the illumination comes, can recommend the desirable photographic exposure for the shot.

And that is what the dome collector on an incident light exposure meter does.

Best regards,

Doug

 

[Doug Kerr]

Well, once I pick up my pen it is hard for me to put it down!

Since I recently announced here a new issue of my article on "Norwood's dome". I have further massaged it a couple of times.

The latest issue as of this writing is Issue 7. The link given in an earlier message to its listing on the index page of The Pumpkin is still valid, but here it is again:

The Pumpkin

Index page of Doug Kerr's The Pumpkin

dougkerr.net

Best regards,

Doug

 

[Doug Kerr]

Don Norwood's original vision, from which the development of the "Norwood's dome" incident light exposure metering system unfolded, starts with this assertion (and I paraphrase):

"As to a photo of a human face, under varied lighting setups, the 'proper' photographic exposure is that based in the average illuminance on the face (the part visible to the camera) on the same basis as if the illumination on the face was wholly 'head on'."

I have poo-pooed this outlook as not being supported by any kind of photometric-photographic model.

He also said that a hemisphere was a fairly good approximation of a human face (that is, the part of it that could be seen by the camera.

Norwood then continued to say that, since an exposure meter with a hemispherical receptor responded to the average illuminance on that receptor, the exposure recommendation of such a meter would be appropriate for various (perhaps all) lighting setups for a human face.

I next note that for some while before the introduction of Norwood's metering method, photographers often used
"duplex metering", in which an incident light meter was used to separately measure the illuminance on the subject from the directions of the two principal lights. Those readings (on an illuminance basis) we averaged, and the result fed into the exposure meter's exposure calculator. Its result was considered the proper photographic exposure for the shot.

This was often found to gave a very desirable result.

Even simpler, as a "rule of thumb" for a setup where the principal illumination came from the side of the subject (at 90° from the camera), photographers were advised to use an exposure one stop greater than would have used with the same light source aimed "head on" to the subject (i.e., from essentially the camera position).

This too was often found to gave a very desirable result (for the cases to which it applied).

Back now to Norwood's vision.

It then turns out that, for lighting from the side (from 90°), the theoretical exposure recommendation from Norwood's system would be the same as that theoretically given, for that lighting setup, by either the duplex metering procedure or the "rule of thumb". Fancy that!

So we are left to wonder whether Don Norwood was prescient or just lucky.

Best regards,

Doug

 

[Ted Cousins]

Here's link to a copy of a Norwood Director instructions:

https://dn790000.ca.archive.org/0/i...eter/Norwood Director Exposure Meter_text.pdf

posted for those unfamiliar with the device ...

 

[Doug Kerr]

Hi, Ted,

Here's link to a copy of a Norwood Director instructions:

https://dn790000.ca.archive.org/0/i...eter/Norwood Director Exposure Meter_text.pdf

posted for those unfamiliar with the device ...

Thanks for that. This is very pertinent.

I think this if for the meter that is often described as the Norwood Director "Model C"

Best regards,

Doug

 

[Ted Cousins]

Don Norwood's original vision, from which the development of the "Norwood's dome" incident light exposure metering system unfolded, starts with this assertion (and I paraphrase):

"As to a photo of a human face, under varied lighting setups, the 'proper' photographic exposure is that based in the average illuminance on the face (the part visible to the camera) on the same basis as if the illumination on the face was wholly 'head on'."<big snip>

I next note that for some while before the introduction of Norwood's metering method, photographers often used "duplex metering", in which an incident light meter was used to separately measure the illuminance on the subject from the directions of the two principal lights. Those readings (on an illuminance basis) we averaged, and the result fed into the exposure meter's exposure calculator. Its result was considered the proper photographic exposure for the shot.

This was often found to gave a very desirable result.<snip>

So we are left to wonder whether Don Norwood was prescient or just lucky.

Best regards,

Doug

Doug, as to "we averaged", how was that done?

(fc(key)+fc(fill))/2 ?

 

[Doug Kerr]

Hi, Ted,

Doug, as to "we averaged", how was that done?

I assume that this means the areal average of the illumnance over the portion of the face that is visible to the camera. (Norwood does not bother us with such subleties).

That is, we could divide the visible surface of the face into numerous infinitesimal regions, and for each, measure the illuminance on that region (recall that this must be done taking into account the orientation of the individual region), then sum those for the entire (visble portion of the) face and divide that by the overall surface area of the (visble portion of the) face.

This makes no presumption about where the overall illumination comes from (what sources, where located, etc.).

The aveaging of the measured illuminance of the key light and that of the fill light (each with respect to a plane perpendicular to the direction to that source from the subject) is part of the "duplex metering" procedure. That is a different matter altogether.

Best regards,

Doug

 

[Doug Kerr]

Hi, Ted,.

Now when we think of the "average illuminance on the henispjherical collector of a Norwood-style exposure meter the same concept applies.

But in my paper, when I evaluated that for some single hypothetical light source and for different orientations of the hemisphere, I did not do that in a way that followed by earlier description. That is, I did not divide the surface of the hemisphere into infinitesimal regions, evaluate the illuminance on each, sum that, and divide the sum by the overall surface area of the dome. Doing so would have involved some tricky spherical trigonometry followed by double integration.

Rather I took an approach that was based on the definition of illuminance itself, which is:

The illuminance on a surface at a point is the amount of luminous flux incident on the surface at that point per unit surface area.

Note that this is independent of the angle from which that flux arrives.

Where the direction of arrival comes into our work is when we consider a beam of a certain luminous flux density striking a surface at some angle. The flux deposited per unit surface area is the luminous flux density times the cosine of the angle of incidence.

This just comes about because of the geometry. A given small increment of actual surface area presents as a smaller projected area to an beam arriving at an angle to "head on". The amount of flux captured from the beam by that region depends on that projected area, where the illuminance which that captured flux creates depends on the actual surface area over which it lands.

But if we start (as in the definition of illuminance) with the amount of luminous flux actually striking some region of the surface, that has already been taken care of.

So I reckoned the total luminous flux that would strike the dome in some orientation (again, the angle from which it arrives at each place is not an issue), and divided that by the total surface area of the dome. That is by definition the average illuminance over the dome.

Best regards,

Doug

 

[Ted Cousins]

Hi, Ted,.

Now when we think of the "average illuminance on the henispjherical collector of a Norwood-style exposure meter the same concept applies.

But in my paper, when I evaluated that for some single hypothetical light source and for different orientations of the hemisphere, I did not do that in a way that followed by earlier description. That is, I did not divide the surface of the hemisphere into infinitesimal regions, evaluate the illuminance on each, sum that, and divide the sum by the overall surface area of the dome. Doing so would have involved some tricky spherical trigonometry followed by double integration.

Rather I took an approach that was based on the definition of illuminance itself, which is:

The illuminance on a surface at a point is the amount of luminous flux incident on the surface at that point per unit surface area.

Note that this is independent of the angle from which that flux arrives.

Where the direction of arrival comes into our work is when we consider a beam of a certain luminous flux density striking a surface at some angle. The flux deposited per unit surface area is the luminous flux density times the cosine of the angle of incidence.

This just comes about because of the geometry. A given small increment of actual surface area presents as a smaller projected area to an beam arriving at an angle to "head on". The amount of flux captured from the beam by that region depends on that projected area, where the illuminance which that captured flux creates depends on the actual surface area over which it lands.

But if we start (as in the definition of illuminance) with the amount of luminous flux actually striking some region of the surface, that has already been taken care of.

So I reckoned the total luminous flux that would strike the dome in some orientation (again, the angle from which it arrives at each place is not an issue), and divided that by the total surface area of the dome. That is by definition the average illuminance over the dome.

Best regards,

Doug

Thanks, I only just now discovered these comprehensive responses. I was curious about the photographer simply pointing his device at the key light with the fill light off and then at the fill light with the key light off, and was just referring to your " ... Those readings (on an illuminance basis) [were] averaged, and the result fed into the exposure meter's exposure calculator.".

Say he gets 500 fc and 200 fc respectively ...

1) how does he "average" those readings in the field, and

2) how is that "average" applied to his light meter to get an exposure recommendation?

All the best,

Ted

 

[Doug Kerr]

Hi, Ted,

Thanks, I only just now discovered these comprehensive responses. I was curious about the photographer simply pointing his device at the key light with the fill light off and then at the fill light with the key light off.

Say he gets 500 fc and 200 fc respectively ...

1) how does he "average" those results in the field, and

The average is (500+200)/2=350 fc

2) how is that "average" applied to his light meter to get an exposure recommendation?

We assume that the way (in simple use) the meter movement reading is entered into the exposure calculator is from the "footcandle" scale on the meter movement. (True, for example, on the Sekonic L-398A.)

In that case, in duplex metering, the calculated average of the two illuminance readings (e.g., 350 fc, per the example above) is entered into the footcandle input on the exposure calculator.

In many exposure meters, the reading of the meter movement is transferred into the exposure calculator on the basis of an arbitrary numerical scale, or even by a "match the needle position" process. Such exposure meters do not lend themselves to duplex metering.

Best regards,

Doug

 

[Doug Kerr]

Hi, Ted,

The "instructions typically given for "duplex metering" often do not distinguish between:

• Separately measuring the luminance with the meter facing the key light and then the fill light, the other source being off at the time. and

• Separately measuring the luminance with the meter facing the key light and then the fill light, both being lit at the time.

It turns out that in the important and interesting case of the key light being at 90° to the camera position, and the fill light being essentially at the camera position, these would give the same result (since the response of the meter in the luminance configuration would be expected to have a directivity of 0 at 90°, so it would ignore the "other" source, if lit, during each of the two measurements).

Best regards,

Doug

 

[Doug Kerr]

Hi, Ted,

Here I will ruminate a bit on Norwood's original vision, which I paraphrase as, "The proper exposure for a subject's face illuminated by one or more light sources is determined from the average illuminance over the part of the face visible to the camera." I have suggested that there is no "theoretical" premise for this.

It is to me reasonable to assume that this means the exposure recommendation we would get by entering that "average illuminance" value into the form of the "standard incident light exposure metering equation" that we otherwise would favor.

In his later paper, Norwood says that the "effectivity" of a certain light source (we may assume that this means one having a certain luminous flux density to its beam as it arrives at the subject), in actually creating a certain "exposure result", varies with the angle from which the source's beam arrives.

Norwood, in his series of subjective experiments, ascertained this effectivity as a function of the angle of the light source. In these tests, viewer subjects were asked to, for each of several light source angles, compare the "exposure result" of shots of various human faces, taken with various photographic exposures, to a "reference" shot taken with the lighting "head on" (which again we might reasonably assume was made with the photographic exposure recommended on the basis of the illuminance from the light source using some acceptable version of the incident light exposure metering equation).

A point on this curve that is of special interest is for a light source angle of 90° (that is, with the light coming "exactly from one side or the other").

For this we have:

• In Norwood's subjective tests, the average "effectivity" of such a light source was 50% of the "effectivity" of the same light source at an angle of 0 (that is, coming from the camera position).

• Norwood suggests that a hemisphere is a reasonable proxy for a subject's face. The average illuminance on such a hemisphere from a light source at 90° is theoretically exactly 50% of the average illuminance on such a hemisphere from the same light source at an angle of 0.

We find at least approximately that same agreement for other light source angles.

That all seems to vindicate Norwood's original "vision".

Best regards,

Doug

 

[Ted Cousins]

Hi, Ted,


The average is (500+200)/2=350 fc

We assume that the way (in simple use) the meter movement reading is entered into the exposure calculator is from the "footcandle" scale on the meter movement. (True, for example, on the Sekonic L-398A.)

In that case, in duplex metering, the calculated average of the two illuminance readings (e.g., 350 fc, per the example above) is entered into the footcandle input on the exposure calculator.

Got it, thanks!

In many exposure meters, the reading of the meter movement is transferred into the exposure calculator on the basis of an arbitrary numerical scale, or even by a "match the needle position" process. Such exposure meters do not lend themselves to duplex metering.

Yes, I have one of those - a Sekonic Auto-Lum "match the needle". It does have a tiny EV window to think about ...

best,

Ted.

 

[Ted Cousins]

Got it, thanks!

Yes, I have one of those - a Sekonic Auto-Lum "match the needle". It does have a tiny EV window to think about ...

... for illuminance EV = log2(ExS/C) - not sure how that works for averaging.

If I get log2(500x10.764 lx x 100/250) = 11.1 EV and
if I get log2(200x10.764 lx x 100/250) = 9.8 EV, both on the calculator dial

what would represent 350 fc? 10.6 EV?


best,

Ted.

 

[Doug Kerr]

Hi, Ted,

I had suggested a while ago that in "duplex metering" the two meter readings, in foot candles, were to be averaged (arithmetically) and the resulting average set into the calculator to get the exposure recommendation.

I have just reviewed the "instructions" for this metering procedure approach in the manuals for several Sekonic "Norwood Director" style exposure meters. I note that seemingly in the manuals for the "Norwood Director" family of meter, duplex metering is not suggested for portraiit work (because of course the dome takes care of that).

I the manuals for several of the later versions, essentially this technique is recommended for certain "landscape" situations One measurement is made with the instrument "with the [meter] at the camera position"*, and the second with the meter pointed at the sun. Both measurements were to be made with the dome light collector in place.

*It is not clear what that should mean. Incident light measurement is ordinarily measured at the subject location. For landscape work, though, the meter is often placed at the camera location, perhaps facing toward the camera.

In the manual for the Sekonic L-28c exposure meter, the instructions for this process start by describing what turns out to the be the geometric (not arithmetic) average of the two footcandle readings, that to be entered into the calculator to give the photographic exposure recommendation.

But it continues to give an alternate procedure. There, for each reading, a recommended shutter speed is noted (for the aperture that will be used), and the "average" of those shutter speeds used.

But is it not clear what the "average" of two shutter speeds means. These speeds are often seen as the denominator" of the speed (for example,. we see "250" for a speed of 1/250 sec).

So perhaps what is meant is, if the two measurements suggest shutter speeds of 1/125 and 1/250, the speed to be used is 1/188 sec.

This is not the same result that would come from the use of the geometric mean of the two footcandle readings, but for the values likely encountered, there is not a great difference. And in any case this is not an exact science

So, go figger!

In the manual for the Sekonic L-398 exposure meter, the second form of the procedure is not mentioned. Same for the manual for the next version, the L-398M.

In the manual for the Sekonic L-398A exposure meter, (later yet), this whole notion is not mentioned.

Best regards,

Doug