Exposure metering and careless language

[Doug Kerr]

In dealing with the matter of exposure metering (as practiced by free-standing exposure meters), a great deal of trouble is caused by expressions that have become part of the "language" of the field that are not really "incorrect" but in fact do not really say something specific (that could be correct or not).

One favorite is: "Exposure meters are generally calibrated to a scene reflectance of 18%"

There is of course a rather convoluted story that involves reflected light exposure metering, a widely adopted (but rarely articulated) "exposure strategy", and an assumed scene average reflectance of 18%. But the statement above, which tries to encapsulate that story in a single sentence, cannot.

Another favorite is: "The result [of a metered exposure] will be [on the sensor] an image that is average gray".

Now does that mean, "The average phtometric exposure on the sensor is what we can consider 'gray' "?

What gray would that be?

Or does it mean, "The phtometric exposure on the sensor is what we can consider 'average gray' "?

The photometric exposure at what point of the image on the sensor?

But this is what is sort of true: ""The average phtometric exposure on the sensor, considered as a fraction of the saturation phtometric exposure, is what we can consider a 'mid gray' "?

(This is rationalized based on the scale of the nonlinear luminance component L*, of the L*a*b* color space.)

Best regards,

Doug

 

[Doug Kerr]

Hi, Ted,

Again I got an email, with a reply from you to my post on this matter,, but I don't see it on the forum!

In any case, as to your Foveon data, I haven't yet followed the train of how that was produced, so I can't really know what to expect.

I any case, i am really not big on the notion of "mid gray" and what is it and what isn't it!

But in any case, remember, when we try and describe the expected photometric exposure on the sensor in whatever terms, we are ordinarily doing that for a (real or hypothetical) metered exposure (and I try to be very explicit about that),

In the case of what is quite reasonably considered a "greatly underexposed" image, we cannot expect any "iconic" benchmarks of phtometric exposure on the sensor to have been attained.

Best regards,

Doug

 

[Doug Kerr]

Hi, Ted,

In any case, if we use reflected light metering, and make a metered shot of some scene with a meter with K=12.6 (and for simplicity's sake I will assume the lens transmittance to be 1), then the average phtometric exposure on the sensor will be essentially 0.128 Hsat.

There is a rationale for calling this "mid gray": if that as a fraction of the maximum value is encoded into the L*a*b* color space, L* (the encoded nonlinear luminance) comes out 50 on a scale of 0-100.

I do not call it "mid gray".

This is of course all more than enough to makes one's head hurt!

Best regards,

Doug

 

[Ted Cousins]

Hi, Ted,

Again I got an email, with a reply from you to my post on this matter,, but I don't see it on the forum!

Yesterday deleted two responses having realized that I may have misunderstood your text. Maybe Openphotography notified you automatically as to what I said since you do mention Foveon data.

In any case, as to your Foveon data, I haven't yet followed the train of how that was produced, so I can't really know what to expect.

If it was a three-channel underexposed raw histogram, consider it withdrawn.

I any case, i am really not big on the notion of "mid gray" and what is it and what isn't it!

But in any case, remember, when we try and describe the expected photometric exposure on the sensor in whatever terms, we are ordinarily doing that for a (real or hypothetical) metered exposure (and I try to be very explicit about that),

I produce raw histograms with RawDigger. For Foveon sensors, it obviously produces a three-channel plot, there being no Bayer or other CFA to contend with.

The withdrawn histogram was from a real metered exposure.

In the case of what is quite reasonably considered a "greatly underexposed" image, we cannot expect any "iconic" benchmarks of [photometric] exposure on the sensor to have been attained.

It was from a series of real metered series of seven shots of a Kodak R27 gray card, -3 to 3 EV in 1 EV steps. I decided that they were not relevant to what you said in this thread: "this is what is sort of true: ""The average phtometric exposure on the sensor, considered as a fraction of the saturation phtometric exposure, is what we can consider a 'mid gray' "?"
.

So, Doug, at what over/underexposure level is an "iconic" benchmark attained and what is an "Iconic benchmark"?

Best regards,

Ted

 

[Doug Kerr]

Hi, Ted,

Yesterday deleted two responses having realized that I may have misunderstood your text.

Yeah, I just figgered that out!

Maybe Openphotography notified you automatically as to what I said since you do mention Foveon data.

No, it's because I had tagged that thread for me to get an email notice of any traffic on it.

So, Doug, at what over/underexposure level is an "iconic" benchmark attained and what is an "Iconic benchmark"?

I probably should not have used that term!

But what I probably had in mind was a benchmark like:

"The photometric exposure on the sensor from a (possibly hypothetical) scene element with R=100% should be [some fraction of] the saturation phtometric exposure."

Of course, for incident light metering, ISO 12232 and ISO 2720 don't work together to attain that since the meter manufacturer can use any value of C they wish (over a modest range).

And for reflected light metering, ISO 12232 and ISO 2720 don't work together to attain that since:

• The meter manufacturer can use any value of K they wish (over a modest range).

• The result depends on the average reflectance of the scene

Why the latter dependency? Well, I have postulated an "exposure objective" in terms of the phtometric exposure of the "brightest" element of the scene, and to get that, we kneed to know the luminance of such an element. But a reflected light exposure meter perceives the average luminance of the scene, which (compared to the maximum possible luminance) depends on the average scene reflectance.

"Overexposure" and "underexposure" only have meaning once we have adopted a certain "exposure objective".

Best regards,

Doug

 

[Ted Cousins]

Hi, Ted,

Yeah, I just figgered that out!

No, it's because I had tagged that thread for me to get an email notice of any traffic on it.

• The result depends on the average reflectance of the scene

Why the latter dependency? Well, I have postulated an "exposure objective" in terms of the [photometric] exposure of the "brightest" element of the scene, and to get that, we [need] to know the luminance of such an element, whereas a reflected light exposure meter perceives the average luminance of the scene, which (compared to the maximum possible luminance) depends on the average scene reflectance.

Best regards,

Doug

Interestingly, RawDigger can select any part of a raw capture and produce a histogram from it which can be compared to an already-measured max sensor response.

 

[Doug Kerr]

Hi, Ted,

Interestingly, RawDigger can select any part of a raw capture and produce a histogram from it which can be compared to an already-measured max sensor response.

That sounds quite useful in looking into the "exposure result" of a given shot.

Best regards,

Doug

 

[Ted Cousins]

Oops!

As to the "under-exposed" post we discussed that I had deleted, I was little bothered as to why the output image was not even trying to be gray, so I just now checked the raw EXIF in RawDigger and found that I had used Manual exposure for that series, duh ...

rgds,

Ted.

 

[Doug Kerr]

Hi, Ted,

As to the "under-exposed" post we discussed that I had deleted, I was little bothered as to why the output image was not even trying to be gray, so I just now checked the raw EXIF in RawDigger and found that I had used Manual exposure for that series, duh ...

Easy to lose track of just what has happened and how!

Best regards,

Doug