Doug Kerr
Well-known member
The motivation for incident light exposure metering
The motivation for incident light exposure metering, simplistically, is this: when shooting a scene comprising a white car on a snowbank, we would like the image to look like a white cat on a snowdrift, not (as we might get with reflected light metering) like a gray cat on an ash heap.
In technical terms
One technical definition of a result of that general type is this: each element of the scene should receives a photometric exposure which, as a fraction of the saturation photometric exposure, corresponds to the reflectance of the element. An element with a reflectance of 35% should received a photometric exposure of 35% of saturation. (Not all will agree that this should be our precise objective. You are welcome to do the algebra that fits the definition you prefer.)
Incident light metering
An incident light exposure meter measures the illuminance on the scene and, modeling in one equation a long trail of photometric algebra, gives us a recommended photographic exposure (combination of shutter speed and aperture) that we hope will bring the result described above.
The "metering equation" used by the meter for this reckoning is scaled by the constant C, the incident light exposure metering calibration constant. The international standard for such meters admits a substantial range for this constant. Some important modern incident light exposure meters use a value in the general area of 340.
If we assume that:
• the value of C for the meter is in fact 340,
• the camera's exposure metering system uses as its exposure index the ISO speed of the sensor (not an alternative metric such as ISO SOS), and
• the transmission factor of the lens is 0.92 (why not)
and we go through the photometric algebra, we find out that we should expect on the sensor for each scene element a photometric exposure that indeed is the fraction of the saturation photometric exposure that matches the reflectance of the element.
How about that!
Doing it without a meter
If we do not have an incident light exposure meter, we can emulate its function by having our camera's inbuilt exposure metering system (a reflected light system) regard a diffuse-reflecting "gray card" of known reflectance placed in the scene where it will be illuminated by the incident light we expect the subject to receive.
If we have these conditions:
• The reflectance of the gray card is 18% (common, for historical reasons).
• The value of K (the reflected light exposure metering calibration constant) for the inbuilt meter is 12.5 (quite common), and the metering is "through the lens" (so the lens transmission factor is automatically compensated for and thus need not be considered).
• the camera's exposure metering system uses as its exposure index the ISO speed of the sensor (not an alternative metric such as ISO SOS), and
and we capture the photographic exposure the camera would use for a "metered" shot of the gray card, and use that for the actual shot of the scene, we would expect on the sensor for each scene element a photometric exposure that is, as the fraction of the saturation photometric exposure, about 0.71 of the reflectance of the element.
So this "emulation" of an actual incident light meter leads to an exposure about 1/2 stop shy of that which would be recommended by an actual incident light exposure meter.
So to follow the same strategy as we would be given by the actual reflected-light exposure meter, we would need to "bump" the exposure chosen by the camera when regarding the gray card by about 1/2 stop to get the exposure to be used for the actual shot.
Many modern digital cameras (including many Canon models) describe the sensitivity of the sensor (for each of several possible settings) in terms of the metric ISO SOS (standard output sensitivity), rather than in terms of ISO speed. The ISO SOS is consistently about 0.71 of the ISO speed for the same sensor sensitivity.
If that metric is the one we set into the incident light exposure meter, or is automatically set as the exposure index in our camera's metering system, then the exposures resulting from the metering (direct or emulated) will be about 1/2 stop "hotter" than I mentioned above.
In the case of metering with an actual incident light exposure meter (with C=340), that exposure could cause clipping of highlights.
In the case of the emulated metering using the camera and a gray card, the resulting exposure will probably be "just about right", and so no "bump" in exposure will typically be needed.
Best regards,
Doug
The motivation for incident light exposure metering, simplistically, is this: when shooting a scene comprising a white car on a snowbank, we would like the image to look like a white cat on a snowdrift, not (as we might get with reflected light metering) like a gray cat on an ash heap.
In technical terms
One technical definition of a result of that general type is this: each element of the scene should receives a photometric exposure which, as a fraction of the saturation photometric exposure, corresponds to the reflectance of the element. An element with a reflectance of 35% should received a photometric exposure of 35% of saturation. (Not all will agree that this should be our precise objective. You are welcome to do the algebra that fits the definition you prefer.)
Incident light metering
An incident light exposure meter measures the illuminance on the scene and, modeling in one equation a long trail of photometric algebra, gives us a recommended photographic exposure (combination of shutter speed and aperture) that we hope will bring the result described above.
The "metering equation" used by the meter for this reckoning is scaled by the constant C, the incident light exposure metering calibration constant. The international standard for such meters admits a substantial range for this constant. Some important modern incident light exposure meters use a value in the general area of 340.
If we assume that:
• the value of C for the meter is in fact 340,
• the camera's exposure metering system uses as its exposure index the ISO speed of the sensor (not an alternative metric such as ISO SOS), and
• the transmission factor of the lens is 0.92 (why not)
and we go through the photometric algebra, we find out that we should expect on the sensor for each scene element a photometric exposure that indeed is the fraction of the saturation photometric exposure that matches the reflectance of the element.
How about that!
Doing it without a meter
If we do not have an incident light exposure meter, we can emulate its function by having our camera's inbuilt exposure metering system (a reflected light system) regard a diffuse-reflecting "gray card" of known reflectance placed in the scene where it will be illuminated by the incident light we expect the subject to receive.
If we have these conditions:
• The reflectance of the gray card is 18% (common, for historical reasons).
• The value of K (the reflected light exposure metering calibration constant) for the inbuilt meter is 12.5 (quite common), and the metering is "through the lens" (so the lens transmission factor is automatically compensated for and thus need not be considered).
• the camera's exposure metering system uses as its exposure index the ISO speed of the sensor (not an alternative metric such as ISO SOS), and
and we capture the photographic exposure the camera would use for a "metered" shot of the gray card, and use that for the actual shot of the scene, we would expect on the sensor for each scene element a photometric exposure that is, as the fraction of the saturation photometric exposure, about 0.71 of the reflectance of the element.
So this "emulation" of an actual incident light meter leads to an exposure about 1/2 stop shy of that which would be recommended by an actual incident light exposure meter.
So to follow the same strategy as we would be given by the actual reflected-light exposure meter, we would need to "bump" the exposure chosen by the camera when regarding the gray card by about 1/2 stop to get the exposure to be used for the actual shot.
This is in fact what was recommended at one time by the instructions accompanying the famous Kodak "gray card" (whose reflectance is indeed nominally 18%).
What about the ISO SOS?Many modern digital cameras (including many Canon models) describe the sensitivity of the sensor (for each of several possible settings) in terms of the metric ISO SOS (standard output sensitivity), rather than in terms of ISO speed. The ISO SOS is consistently about 0.71 of the ISO speed for the same sensor sensitivity.
If that metric is the one we set into the incident light exposure meter, or is automatically set as the exposure index in our camera's metering system, then the exposures resulting from the metering (direct or emulated) will be about 1/2 stop "hotter" than I mentioned above.
In the case of metering with an actual incident light exposure meter (with C=340), that exposure could cause clipping of highlights.
In the case of the emulated metering using the camera and a gray card, the resulting exposure will probably be "just about right", and so no "bump" in exposure will typically be needed.
Best regards,
Doug