Hi, Bart,
I know Glenn did, but in my experience it never hurts to double check the obvious, because what is obvious to one may not be obvious to another.
Absolutely, and I was being sarcastic with my observation!
In my experience with a different model Sekonic lightmeter, one needs to adjust the 'Lumisphere' in one of two positions, and they give different readings. One position is with a receded Lumisphere, but that is basically used to simulate a flat receptor, as in a Lux meter, and is used to set up lights (lighting contrast) and measure for exposures of flat surfaces (where one still needs to correct for direct reflection of glossy sufaces).
However, wth the extended Lumisphere setting, one has an incident light meter that handles angles of light like 3D subjects do, shading and all (when used at the subject, pointing towards the camera). The difference between the two methods can give up to something like 1.5 stops difference in suggested exposure.
I hadn't been familiar with those two configurations. I'm trying (as you might imagine) to relate them to different theoretical outlooks (you know, those pesky cosines!).
Your comment about glossy surfaces is, I'm sure, one of the keys - the way we deal with "non-Lambertian" surfaces.
It also doesn't hurt to 'calibrate' the meter to one's camera. A reflected metering from the camera position of a uniformly lit surface should produce a histogram with a spike, approx. centered between the edges. An incident light metering will only give the same result for a surface of appox. 12.5% reflection (which is what my Sekonic is factory calibrated for). Of course one can 'calibrate' the meter for a different response, e.g. by setting the ISO on the meter different to the camera's ISO setting.
And indeed the relationships you mention vary with two parameters of the camera's automatic exposure control system:
• What we can think of as the "calibration" of the exposure metering system itself.
• How the manufacturer assesses the ISO sensitivity of the sensor system.
The ISO standard for (free-standing) exposure meters gives a wide range of "calibration factors" that are considered acceptable. The standard for integrated automatic exposure systems does not (there is of course a tolerance expressed, but not a range of "acceptable nominal values").
If we take the calibration specified by the latter standard, and convert it to be applicable to a free-standing exposure meter (reflected-light mode), and postulate:
• A free-standing exposure meter calibrated to that norm (which is typical for serious professional exposure meters), and
• A camera in which the ISO sensitivities are assessed in accordance with the original prescription of the applicable ISO "sensitivity" standard for digital cameras (the "ISO speed" measure)
then we can theoretically expect an exposure, guided by the exposure meter setting (reflected light mode), of a uniform-luminance scene (I choose that just to minimize any complication in what is meant), to give an image with RGB coordinates that represent a relative luminance of about 12.8% of the "saturation" value.
On the other hand, if the ISO sensitivity of the camera is assessed in accordance with the new alternate measure of sensitivity defined by the current applicable ISO standard (the "ISO SOS" measure - Standard Output Sensitivity), then we can expect the exposure described above to give an image with RGB coordinates that represent about 18% of the "saturation" photometric exposure.
Modern Canon EOS cameras evidently have ISO sensitivity ratings assessed in term of the "SOS" measure (and are thus, numerically, about 70% of the value that would be assessed under the original "ISO speed" measure).
An interesting corollary is that on my EOS 40D, midscale on the in-camera histogram corresponds very nearly (for a "neutral" object) to RGB values representing a photometric exposure of 18% of saturation luminance.
To finish the Canon story, Canon these days, although evidently using the "SOS" premise to determine their ISO "sensitivity" values, chooses to express them in terms of a second "new" alternative measure, the "ISO REI" (Recommended Exposure Index). Its definition is this (I have paraphrased to most clearly express the concept):
The value that, set into an exposure metering system as the exposure index, will produce a desirable exposure result for many users in many cases.
Of course, the advantage of expressing the sensitivities in that fashion is that nobody can make measurements of actual cameras and then challenge the "truthfulness" of Canon's published "ISO" numbers.
Thanks so much for your input into this.
Best regards,
Doug