Doug Kerr
Well-known member
We often hear that "exposure meters are (should be) calibrated to 18%." (Sometimes we hear, "18% gray".)
Of course, that doesn't describe any technical arrangement. But it is a "code phase" for a certain situation; unfortunately most people don't really know what it is.
I'll talk a little about that.
**********
Reflected light metering
A reflected light exposure meter (or the "metering" portion of an integrated automatic exposure control system) recommends, or sets, an exposure (in the sense of an aperture and shutter speed) using an equation with two inputs:
• The measured average luminance of the scene
• The exposure index, which we can think of as being what we (or the camera) tells the exposure meter is the ISO sensitivity of the film or digital imaging chain.
The exact way that equation works can be said to be the calibration of the exposure meter.
International standards
International standard ISO 2721 prescribes a recommended calibration for integrated exposure control systems.
International standard ISO 12232 defines how to determine the "ISO speed" (a measure of sensitivity) of a digital camera.
Their implication
If we have a camera:
• The metering portion of whose exposure control system follows ISO 2721, and
• In which the assignment of "ISO speed" values for the various "ISO" choices follows ISO 12232
then we should have a situation that can be described in two ways:
• If we take a metered exposure of a uniform-luminance scene, the photometric exposure (that's the product of illuminance and exposure time) on the sensor should everywhere be 12.8% of the saturation photometric exposure, which is defined as the photometric exposure that would (just) lead to the highest possible digital representation in the "developed" image.
• If we take a metered exposure of a scene, the average photometric exposure on the sensor should be 12.8% of the saturation photometric exposure.
This in turn could be looked at in two ways (among others):
• If we have a uniformly-illuminated scene whose average reflectance is 12.8%, then an object in the scene with a reflectance of 100% will have a photometric exposure in the image that is just at saturation.
• If we have a uniformly-illuminated scene whose average reflectance is 18%, then an object in the scene with a reflectance of 100% will have a photometric exposure in the image that is 1/2 stop down" from saturation.
So that's where the "18%" number comes from. It's the average reflectance of a scene that would give us "1/2 stop headroom" with respect to any 100% reflectance objects in the scene.
Canon cameras
Over the years, Canon cameras (especially the dSLRs) have produced an exposure almost 1/2 stop "hotter" than what was described above. They evidently found that this gives a good result in many cases.
They could have done this in either of these ways:
a. Making the calibration of the exposure meter part of the automatic exposure control system "1/2 stop hotter" than that recommended by ISO 2721.
b. Reckoning of the ISO sensitivity of the camera at about 71% that which would be determined per ISO 12232 ("1/2 stop understated").
Had they done (a), then users with external exposure meters would have noticed a discrepancy between the exposures they recommended and those enacted by the camera's automatic exposure control system, and that would have caused some consternation. So they did (b).
New ISO measures
Based on recommendations by the Camera and Imaging Products Association (Japan) (CIPA), the latest edition of ISO 12232 (ISO 12232-2006) introduces two alternative measures of digital camera sensitivity:
• The ISO standard output sensitivity (SOS). This is for all practical purposes defined in the same way as the ISO speed defined by ISO 12232, except that the value is 0.713 times the ISO speed (essentially "1/2 stop" lower).
A rationale for this is that modern exposure metering systems, being more sophisticated than "scene average metering systems", makes the "1/2 stop headroom" unnecessary.
Of course we only had 1/2 stop headroom for a scene whose average reflectance was 18%. But if you believe that 18% average reflectance is somehow representative of real scenes, you'll believe that there was 1/2 stop of headroom.
• The ISO recommended exposure index (REI). This is the exposure index the camera manufacturer recommends be used in making exposure determinations. There is no objective basis for its determination. The assumption is that the manufacturer would establish it based on extensive empirical testing, perhaps with user preference rating. And nobody could accuse a manufacturer's values of being "wrong".
Now, about Canon
Firstly, we can say that the introduction of the ISO SOS measure essentially "legitimatizes" Canon's practice all the years regarding the matter we discussed here. That is, Canon has been very nearly rating the "ISO sensitivities" of their cameras as defined for the ISO SOS measure.
But, rather than officially adopting the ISO SOS, rather than the traditional "ISO speed" measure, as the premise of their ISO ratings, they have decided to "advertise" the sensitivity of their cameras as being on the basis of the ISO REI. Now, they can't possibly be "wrong".
Of course, that doesn't describe any technical arrangement. But it is a "code phase" for a certain situation; unfortunately most people don't really know what it is.
I'll talk a little about that.
**********
Reflected light metering
A reflected light exposure meter (or the "metering" portion of an integrated automatic exposure control system) recommends, or sets, an exposure (in the sense of an aperture and shutter speed) using an equation with two inputs:
• The measured average luminance of the scene
• The exposure index, which we can think of as being what we (or the camera) tells the exposure meter is the ISO sensitivity of the film or digital imaging chain.
The exact way that equation works can be said to be the calibration of the exposure meter.
International standards
International standard ISO 2721 prescribes a recommended calibration for integrated exposure control systems.
International standard ISO 12232 defines how to determine the "ISO speed" (a measure of sensitivity) of a digital camera.
Their implication
If we have a camera:
• The metering portion of whose exposure control system follows ISO 2721, and
• In which the assignment of "ISO speed" values for the various "ISO" choices follows ISO 12232
then we should have a situation that can be described in two ways:
• If we take a metered exposure of a uniform-luminance scene, the photometric exposure (that's the product of illuminance and exposure time) on the sensor should everywhere be 12.8% of the saturation photometric exposure, which is defined as the photometric exposure that would (just) lead to the highest possible digital representation in the "developed" image.
• If we take a metered exposure of a scene, the average photometric exposure on the sensor should be 12.8% of the saturation photometric exposure.
This in turn could be looked at in two ways (among others):
• If we have a uniformly-illuminated scene whose average reflectance is 12.8%, then an object in the scene with a reflectance of 100% will have a photometric exposure in the image that is just at saturation.
• If we have a uniformly-illuminated scene whose average reflectance is 18%, then an object in the scene with a reflectance of 100% will have a photometric exposure in the image that is 1/2 stop down" from saturation.
So that's where the "18%" number comes from. It's the average reflectance of a scene that would give us "1/2 stop headroom" with respect to any 100% reflectance objects in the scene.
Is that situation well described as "18% calibration of the exposure meter"?
Canon cameras
Over the years, Canon cameras (especially the dSLRs) have produced an exposure almost 1/2 stop "hotter" than what was described above. They evidently found that this gives a good result in many cases.
They could have done this in either of these ways:
a. Making the calibration of the exposure meter part of the automatic exposure control system "1/2 stop hotter" than that recommended by ISO 2721.
b. Reckoning of the ISO sensitivity of the camera at about 71% that which would be determined per ISO 12232 ("1/2 stop understated").
Had they done (a), then users with external exposure meters would have noticed a discrepancy between the exposures they recommended and those enacted by the camera's automatic exposure control system, and that would have caused some consternation. So they did (b).
New ISO measures
Based on recommendations by the Camera and Imaging Products Association (Japan) (CIPA), the latest edition of ISO 12232 (ISO 12232-2006) introduces two alternative measures of digital camera sensitivity:
• The ISO standard output sensitivity (SOS). This is for all practical purposes defined in the same way as the ISO speed defined by ISO 12232, except that the value is 0.713 times the ISO speed (essentially "1/2 stop" lower).
A rationale for this is that modern exposure metering systems, being more sophisticated than "scene average metering systems", makes the "1/2 stop headroom" unnecessary.
Of course we only had 1/2 stop headroom for a scene whose average reflectance was 18%. But if you believe that 18% average reflectance is somehow representative of real scenes, you'll believe that there was 1/2 stop of headroom.
• The ISO recommended exposure index (REI). This is the exposure index the camera manufacturer recommends be used in making exposure determinations. There is no objective basis for its determination. The assumption is that the manufacturer would establish it based on extensive empirical testing, perhaps with user preference rating. And nobody could accuse a manufacturer's values of being "wrong".
Now, about Canon
Firstly, we can say that the introduction of the ISO SOS measure essentially "legitimatizes" Canon's practice all the years regarding the matter we discussed here. That is, Canon has been very nearly rating the "ISO sensitivities" of their cameras as defined for the ISO SOS measure.
But, rather than officially adopting the ISO SOS, rather than the traditional "ISO speed" measure, as the premise of their ISO ratings, they have decided to "advertise" the sensitivity of their cameras as being on the basis of the ISO REI. Now, they can't possibly be "wrong".