Doug Kerr
Well-known member
In another thread in this forum section, Fotis D. Tirokomos has provided an extensive monograph on an important topic related to photographic exposure metering. I haven't read it yet, but I plan to soon, and expect to comment on it.
In my comments, I will almost certainly refer to what I call the "fundamental premise of single-valued exposure metering". The actual "premise" is devilishly simple, but the background behind it is a little complicated, and is widely misunderstood.
So I thought I would review it here, sort of like an "include" file in a computer program with respect to my expected comments on Fotis' monograph. And of course it is hoped that this information will be of interest in its own right.
Single-valued exposure metering
"Single valued" exposure metering refers to the situation in which the "measurement organ" returns a single value from its examination of the scene or the lighting on the scene, which is the sole "measurement" input to the exposure metering algorithm. This value may in fact be directly derived with a single overall measurement, averaging the luminance "uniformly" or in accordance with some "weighting" pattern. Or it may be derived by taking measurements at a number of points across the scene, and then averaging them, perhaps uniformly, or perhaps in accordance with some static weighting pattern.
This is as distinguished from systems in which measurements taken across the scene are separate inputs to the algorithm, which then considers them in a way that may include pattern recognition and the like. This approach is widely used in automatic exposure schemes in modern cameras.
But this is not the paradigm I will be treating here. In fact, the "single-valued" class of exposure metering I do treat is what is found almost universally in "free standing" exposure meters, and the scenarios I describe are best understood in the context of free-standing exposure meters.
Two metering techniques
Now, that all having been said, I will consider two significantly-different techniques of exposure metering. They require different instruments (or different configurations of a dual-purpose instrument).
Before I proceed, I will note that there are two situations of scene lighting that may call for different exposure metering techniques. One is continuous lighting (as from the sun or from incandescent or fluorescent lighting). The other is burst lighting, as we have with illumination from a photoflash unit. Both of the classes of exposure metering have forms for both those circumstances. For conciseness of presentation, in general, my base language will be fore the continuous lighting situation, and I will put in square brackets the alternative language applicable to the burst ("flash") lighting situation.
Now, to the two techniques of exposure metering.
"Reflected light" metering systems. Here, the measuring organ determines the luminance [luminance-time product] of the scene over some field of view. Ideally, that field of view would be coterminous with the field of view of the camera for the shot to me made. But a free-standing exposure meter does not in general have any idea what that is, so its field of view may be arbitrary. But I assume here that the field of view of the meter is not "very small", as in the case of "spot" metering patterns.
The single value may be a uniform continuous average over the meter's field of view, a weighted continuous average, or uniform or weighted averages of multiple discrete measurements. The principle I discuss here can be best comprehended if we visualize a continuous true average.
The term"reflected light" metering is not fully apt, since this modality applies equally well to self-luminous objects (backlit signboards, for example).
"incident light" metering systems. Here, the measuring organ determines the illuminance [illuminance-time product] of the illumination on some location in the scene.
The fundamental premise - "reflected light" metering
In "reflected light" metering systems, the metering algorithm seeks to guide our setting of the camera photographic exposure (aperture and shutter speed [aperture only for flash]) to produce a predetermined value of the average photometric exposure on the film or digital sensor.
Now, in any particular meter, that predetermined value of this desired average photometric exposure varies linearly with the exposure index, which is what we tell the meter is the ISO speed (a specific objectively defined measure of sensitivity) of the film or sensor system.
When we set the little dial on a traditional analog free-standing exposure meter to "ISO 400", we are setting the exposure index to that value. This may or may not be the actual sensitivity (in terms of the "ISO speed") of the film in use or the sensor in its current "ISO" mode; the meter of course has no idea what that actually is.
But with the dial set to that value, the algorithm built into the meter will advise us of photographic exposure combinations that will produce, on the film or sensor, an average photometric exposure that the manufacturer of the meter feels is desirable if the ISO speed of the film or sensor were actually ISO 400.
So the result from the measuring organ is the input to the exposure metering algorithm, and the exposure index setting is a parameter. There is another parameter, the exposure meter calibration constant. It is by the choice of this constant that the manufacturer implements his opinion on what is the proper "meter-controlled" average photometric exposure on the film or sensor for any given ISO speed rating (the assumption being that the user will in fact set the exposure index to the rated ISO speed).
What about standards?
Isn't the exposure meter calibration constant prescribed by a international standard?
Yes and no. Yes it is prescribed, but to fall within a "really pretty wide range". The reason for this is historical, going back to the time when each meter manufacturer had his own idea of what "metered expose strategy" he wanted his company's meters to use as their premise for recommending photographic exposure settings to users. As we'll see later, we have not so much moved beyond that.
Incidentally, the definition of the constant (in terms of its unit, for one thing) is different for "reflected light" and "incident light" meters, and thus the nominal numerical values are different as well. In addition, the ranges permitted by the standard (in term of the max/min ratio) are slightly different between the two.
The fundamental premise - "incident light" metering
In "incident light" metering systems, the metering algorithm seeks to guide our setting of the camera exposure (aperture and shutter speed [aperture]) to produce a predetermined value of the average photometric exposure on the film or digital sensor for a certain assumed value of average scene reflectance (and depending on the exposure index setting).
If we consider a reflected light exposure metering system where the exposure meter calibration constant is in the center of the range for that class of meter provided for by the international standard, and consider an incident light exposure metering system where the exposure meter calibration constant is in the center of the range for that class of meter provided for by the international standard, then for any setting of the exposure index, the photometric exposure on the film or sensor resulting from a "metered" exposure will be the same for both meters (both techniques) if the average scene reflectance is about 0.16 (16%). (Perhaps you were expecting 18.3%.)
Integrated exposure metering systems
What about automatic exposure control in cameras? These are integrated exposure metering systems, normally of the "reflected light" type. They are "integrated" with the camera in these ways (in the case of a digital camera):
• When we set the "ISO" to 400, then:
Now, curiously enough, the international standard for automatic exposure control systems, a different standard than for free-standing exposure meters, does prescribe the exposure metering calibration constant with a fairly small range. But it does it in a round-about way, so that the numerical value doesn't really show up as such in the standard (it must be deduced from the actual provisions of the standard)!
Conformity?
Now, if, on a certain camera, when the "ISO" is set to 400, the ISO speed (a specific objectively-defined measure of the photometric sensitivity of a digital sensor) is not ISO 400, isn't that misrepresentation on the part of the manufacturer?
Well, no. If we read the camera specification, we find that the "ISO" setting is not in terms of the objectively-defined property ISO speed, but rather in terms of what is called the ISO Recommended Exposure Index (ISO REI). What is that?
That is a value that the manufacturer is allowed to choose on any basis he desires to, in order that when this value is used as the exposure index in a "standard" automatic exposure metering system it will produce the average photometric exposure that fulfills the "exposure strategy" the manufacturer wants for his cameras' users.
Woof!
Best regards,
Doug
In my comments, I will almost certainly refer to what I call the "fundamental premise of single-valued exposure metering". The actual "premise" is devilishly simple, but the background behind it is a little complicated, and is widely misunderstood.
So I thought I would review it here, sort of like an "include" file in a computer program with respect to my expected comments on Fotis' monograph. And of course it is hoped that this information will be of interest in its own right.
Single-valued exposure metering
"Single valued" exposure metering refers to the situation in which the "measurement organ" returns a single value from its examination of the scene or the lighting on the scene, which is the sole "measurement" input to the exposure metering algorithm. This value may in fact be directly derived with a single overall measurement, averaging the luminance "uniformly" or in accordance with some "weighting" pattern. Or it may be derived by taking measurements at a number of points across the scene, and then averaging them, perhaps uniformly, or perhaps in accordance with some static weighting pattern.
This is as distinguished from systems in which measurements taken across the scene are separate inputs to the algorithm, which then considers them in a way that may include pattern recognition and the like. This approach is widely used in automatic exposure schemes in modern cameras.
But this is not the paradigm I will be treating here. In fact, the "single-valued" class of exposure metering I do treat is what is found almost universally in "free standing" exposure meters, and the scenarios I describe are best understood in the context of free-standing exposure meters.
Two metering techniques
Now, that all having been said, I will consider two significantly-different techniques of exposure metering. They require different instruments (or different configurations of a dual-purpose instrument).
Before I proceed, I will note that there are two situations of scene lighting that may call for different exposure metering techniques. One is continuous lighting (as from the sun or from incandescent or fluorescent lighting). The other is burst lighting, as we have with illumination from a photoflash unit. Both of the classes of exposure metering have forms for both those circumstances. For conciseness of presentation, in general, my base language will be fore the continuous lighting situation, and I will put in square brackets the alternative language applicable to the burst ("flash") lighting situation.
Now, to the two techniques of exposure metering.
"Reflected light" metering systems. Here, the measuring organ determines the luminance [luminance-time product] of the scene over some field of view. Ideally, that field of view would be coterminous with the field of view of the camera for the shot to me made. But a free-standing exposure meter does not in general have any idea what that is, so its field of view may be arbitrary. But I assume here that the field of view of the meter is not "very small", as in the case of "spot" metering patterns.
The single value may be a uniform continuous average over the meter's field of view, a weighted continuous average, or uniform or weighted averages of multiple discrete measurements. The principle I discuss here can be best comprehended if we visualize a continuous true average.
The term"reflected light" metering is not fully apt, since this modality applies equally well to self-luminous objects (backlit signboards, for example).
"incident light" metering systems. Here, the measuring organ determines the illuminance [illuminance-time product] of the illumination on some location in the scene.
The fundamental premise - "reflected light" metering
In "reflected light" metering systems, the metering algorithm seeks to guide our setting of the camera photographic exposure (aperture and shutter speed [aperture only for flash]) to produce a predetermined value of the average photometric exposure on the film or digital sensor.
Photometric exposure is the physical phenomenon to which the film or sensor responds. The quantity is the illuminance-time product on the film or sensor.
Now, in any particular meter, that predetermined value of this desired average photometric exposure varies linearly with the exposure index, which is what we tell the meter is the ISO speed (a specific objectively defined measure of sensitivity) of the film or sensor system.
When we set the little dial on a traditional analog free-standing exposure meter to "ISO 400", we are setting the exposure index to that value. This may or may not be the actual sensitivity (in terms of the "ISO speed") of the film in use or the sensor in its current "ISO" mode; the meter of course has no idea what that actually is.
But with the dial set to that value, the algorithm built into the meter will advise us of photographic exposure combinations that will produce, on the film or sensor, an average photometric exposure that the manufacturer of the meter feels is desirable if the ISO speed of the film or sensor were actually ISO 400.
So the result from the measuring organ is the input to the exposure metering algorithm, and the exposure index setting is a parameter. There is another parameter, the exposure meter calibration constant. It is by the choice of this constant that the manufacturer implements his opinion on what is the proper "meter-controlled" average photometric exposure on the film or sensor for any given ISO speed rating (the assumption being that the user will in fact set the exposure index to the rated ISO speed).
What about standards?
Isn't the exposure meter calibration constant prescribed by a international standard?
Yes and no. Yes it is prescribed, but to fall within a "really pretty wide range". The reason for this is historical, going back to the time when each meter manufacturer had his own idea of what "metered expose strategy" he wanted his company's meters to use as their premise for recommending photographic exposure settings to users. As we'll see later, we have not so much moved beyond that.
Incidentally, the definition of the constant (in terms of its unit, for one thing) is different for "reflected light" and "incident light" meters, and thus the nominal numerical values are different as well. In addition, the ranges permitted by the standard (in term of the max/min ratio) are slightly different between the two.
The fundamental premise - "incident light" metering
In "incident light" metering systems, the metering algorithm seeks to guide our setting of the camera exposure (aperture and shutter speed [aperture]) to produce a predetermined value of the average photometric exposure on the film or digital sensor for a certain assumed value of average scene reflectance (and depending on the exposure index setting).
If we consider a reflected light exposure metering system where the exposure meter calibration constant is in the center of the range for that class of meter provided for by the international standard, and consider an incident light exposure metering system where the exposure meter calibration constant is in the center of the range for that class of meter provided for by the international standard, then for any setting of the exposure index, the photometric exposure on the film or sensor resulting from a "metered" exposure will be the same for both meters (both techniques) if the average scene reflectance is about 0.16 (16%). (Perhaps you were expecting 18.3%.)
Integrated exposure metering systems
What about automatic exposure control in cameras? These are integrated exposure metering systems, normally of the "reflected light" type. They are "integrated" with the camera in these ways (in the case of a digital camera):
• When we set the "ISO" to 400, then:
•• The sensor system is set to a certain sensitivity (I did not say an ISO speed of 400 - more on that in a minute)
•• The exposure index of the exposure metering system is set to ISO 400
• The result of the exposure metering algorithm sets the camera's photographic exposure, or guides us with a "little needle" to set it manually, to produce the average photometric exposure on the sensor the camera manufacturer thinks is appropriate for the sensor sensitivity when the "ISO" is set to 400.•• The exposure index of the exposure metering system is set to ISO 400
Now, curiously enough, the international standard for automatic exposure control systems, a different standard than for free-standing exposure meters, does prescribe the exposure metering calibration constant with a fairly small range. But it does it in a round-about way, so that the numerical value doesn't really show up as such in the standard (it must be deduced from the actual provisions of the standard)!
Conformity?
Now, if, on a certain camera, when the "ISO" is set to 400, the ISO speed (a specific objectively-defined measure of the photometric sensitivity of a digital sensor) is not ISO 400, isn't that misrepresentation on the part of the manufacturer?
Well, no. If we read the camera specification, we find that the "ISO" setting is not in terms of the objectively-defined property ISO speed, but rather in terms of what is called the ISO Recommended Exposure Index (ISO REI). What is that?
That is a value that the manufacturer is allowed to choose on any basis he desires to, in order that when this value is used as the exposure index in a "standard" automatic exposure metering system it will produce the average photometric exposure that fulfills the "exposure strategy" the manufacturer wants for his cameras' users.
Woof!
Best regards,
Doug