• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

A reminder about "ISO sensitivity"

Doug Kerr

Well-known member
Last week, as I contemplated the possibility of adding a new camera or body to my already inflated arsenal, I reviewed the reviews (!) of several reviewing authorities of several candidate cameras.

On one of these sites (sadly, I cannot at the moment reconstruct which one), there was for the various cameras reviews a section on "ISO accuracy", in which a report was given of how something about the camera's sensitivity settings agreed with what it should be.

The simple description of the testing process said it involved the use of a "calibrated" Sekonic L-398A exposure meter (by the way, probably the last of the "Norwood" dynasty of exposure meters), "based on 'medium gray'"

I would be grateful if any of the members here could recognize which review site this is!​

No further details of the test were given (perhaps there are, hidden elsewhere on the site), but I am a little mystified (I might say "worried") about just how the test was done, and what it determined.

Part of my concern comes from the fact that the exposure index setting on the Sekonic L-398A (that's where we tell the meter what it should assume is the ISO speed of the film or digital sensor) is in terms of ISO speed (no kidding), which is one of two common "ISO" measures of digital camera sensitivity.

But in the preponderance of digital cameras today, the "ISO" settings are in terms of the ISO Standard Output Sensitivity (ISO SOS).

Simplistically, for any given digital camera, at any given "ISO" setting, the ISO SOS should be "1/2 stop" lower than the ISO Speed.

Now no doubt the perpetrators of these "ISO accuracy" tests have in fact taken that into account, which may be the reason that for many of the cameras reviewed, the "ISO accuracy", however "they" measured it, was within 1/6 stop of the correct value.

************

By way of review, we might ask why there are these two measures of digital camera sensitivity. The story is not pretty.

The matter of the "calibration" of reflected light exposure meters is a tangled one full of folklore and misdirection. Recall that in a basic reflected light exposure meter, the average luminance of the scene is detected, and a calculation made (in which the exposure index setting is a parameter) to determine the "recommended" photographic exposure.

Now of course in this scenario, the range of photometric exposure on the sensor varies over the scene. The maximum photometric exposure, a critical factor in whether we have "highlight clipping", will vary based on the ratio of maximum to average luxuriance of the scene. If the scene is uniformly illuminated, then (simplistically) that will be the ratio of the maximum reflectivity of areas in the scene to the average reflectivity.

Now, the premise for the "calibration" of a reflected light exposure meter is:

• We assume the maximum reflectivity of areas in the scene to be 100%

• We assume the average reflectivity of areas in the scene to be 18%

Then goes the story, we want the photographic exposure "recommended" by the exposure meter to lead to a maximum photometric exposure on the sensor of 1/2 stop below the saturation photometric exposure (at which clipping occurs).

Woof!

That 1/2 stop is often thought of as a "headroom" (to borrow a term from audio recording) to avert clipping in cases where the assumptions above do not obtain for the actual scene.

Now modern cameras with automatic exposure systems (a coordinated embedment of an exposure meter into the camera) do not simplistically look at the average luminance of the scene, but rather make a determination based on the determination of the luminance in many point in the scene.

However, for continuity, the overall "calibration" of these systems is typically made such that, when presented with a uniform luminance scene (a frame-filling gray test card, perhaps), the automatic photographic exposure setting is consistent with that which would be recommended by a standard reflected light exposure meter.

But it turns out that this approach leads to a too-conservative exposure in many cases. The reason is that 1/2 stop of headroom, which for the classical exposure meter protects against clipping in the case that the assumptions about the distribution of luminance in the scene do not occur. With a "more intelligent" exposure metering scheme, "we don't need no stinkin' headroom".

So various camera manufacturers decided to "eat the headroom". That is, they "bumped" the photographic exposure up by 1/2 stop compared to a metering scheme that matched classical exposure metering calibration (when tested on a uniform illuminate scene).

But there were (at least) three ways to do that:

A. Just bump the "calibration" of the exposure metering system up by 1/2 stop.

The fly in that ointment was that if some over-eager user decided to check the exposure metering system against a well-calibrated exposure meter (yes, I do have a Sekonic L-398A) it would seem that the exposure metering system in the camera was "off" by 1/2 stop, and that would lead to all sorts of garment-rending on dpr, and so forth. So that was not a good plan

B. "Fake" the exposure index.

Suppose the camera's "ISO" setting were 100, and in fact the ISO speed of the imaging system were ISO 100. But we tell the exposure calculating algorithm that the exposure index is 71. Thus we get a 1/2 stop "bump" in the exposure.

But this is actually just the same as scheme "A", looked at in a different way. So it has the same problem.

C. Reckon the ISO sensitivity at 1/2 stop less than its actual value. Thus if we set the ISO sensitivity to 100, the actual ISO speed of the imaging system is ISO 141. But we feed the exposure metering algorithm an exposure index that is exactly the value "set" by the user (ISO 100). Thus we get the desired 1/2 stop "bump" in exposure.

And the eager exposure metering calibration checker will find that, in fact, the calibration of the camera's exposure metering system is "right on". Right on!

But now some testing site, with the means to actually test the sensitivity oi the imaging system, finds out that with the "ISO" setting at 100, the actual sensitivity is ISO 141.

So now there is more garment-rending on dpr. "ISO settings on Canon cameras way wrong."

And in fact Canon labored under this "cloud" for several years.

So eventually various Japanese camera manufacturers, led by Canon, decided that there should be a new, and distinctly-named, measure of the sensitivity of a digital camera which was, essentially, 1/2 stop less than the actual ISO speed.

This was adopted by the ISO committee on camera sensitivity matters, which decided to label this new measure the ISO Standard Output Sensitivity (ISO SOS). And, as I mentioned above, today this is the most widely-used description of digital camera sensitivity. And in fact the standard for EXIF metadata now provides for a "tag" that indicates whether the number in the "ISO sensitivity" tag is the ISO Speed or the ISO SOS. (There is a further possibility, but that is beyond the scope of this note, and doesn't really have anything to do with the issue here.)

Best regards,

Doug
 
One can't help but notice references to DP Review, expressed in lower case diminutively as "dpr".

Not wishing to start a war but may I assume that DPR is not popular here for some reason?
 

Doug Kerr

Well-known member
Hi, Ted,

One can't help but notice references to DP Review, expressed in lower case diminutively as "dpr".

Not wishing to start a war but may I assume that DPR is not popular here for some reason?

Firstly, my use of "dpr" as an initialism of that forum is not meant to be diminutive (as it is when I speak of President Trump). It is just an editorial convention that I for some reason adopted.

But to address your question, as for myself, for a number of years I was a regular participant in the dpr forums. I was regularly showered with abuse. I decided there was no need to expose myself (you should pardon the expression) to that, so I stopped participating.

I don't think that I detect any (visible) general disrespect for the dpr forum here.

i am so pleased that the discourse on OPF is (almost always) civil.

Best regards,

Doug
 
Hi, Ted,

Firstly, my use of "dpr" as an initialism of that forum is not meant to be diminutive (as it is when I speak of President Trump). It is just an editorial convention that I for some reason adopted.

Thanks for the clarification, Doug!

But to address your question, as for myself, for a number of years I was a regular participant in the dpr forums. I was regularly showered with abuse. I decided there was no need to expose myself (you should pardon the expression) to that, so I stopped participating.

My understanding is that some fora there are worse than others. I am a regular on the Sigma cameras forum and less so on their technical forum. Neither of those get too bad, IMHO.

I don't think that I detect any (visible) general disrespect for the dpr forum here.

i am so pleased that the discourse on OPF is (almost always) civil.

With Gents like us on board, how could it be otherwise? ;-)
 

Doug Kerr

Well-known member
Hi, Ted,

Thanks for the clarification, Doug!

My understanding is that some fora there are worse than others. I am a regular on the Sigma cameras forum and less so on their technical forum. Neither of those get too bad, IMHO.

With Gents like us on board, how could it be otherwise? ;-)

Quite!

Best regards,

Doug
 

Doug Kerr

Well-known member
Hi, Ted,
My understanding is that some fora there are worse than others. I am a regular on the Sigma cameras forum and less so on their technical forum. Neither of those get too bad, IMHO.

Yes, each arena is different. I am a member of one group dealing with a very complex technical area with profound historical implications. The members there are so polite that we can imagine them putting on white gloves before taking to the keyboard.

I have been active on another forum dedicated to an important, complex, and well respected software product for a specialized field. The members there are always polite, sometimes perhaps technically condescending, but not often or much.

The moderator, however (he is also the program developer, an incredibly productive fellow) is rude and insulting (although the tone of this waxes and wanes a little bit). Forum messages he doesn't like, or that he thinks are not worthwhile, just disappear, sometimes (forum-wise) their authors as well. I am for example one of "los desaparecidos". Still, when I look in the mirror . . .

A chacun son égout!

Best regards,

Doug
 
Hi, Ted,

I have been active on another forum dedicated to an important, complex, and well respected software product for a specialized field. The members there are always polite, sometimes perhaps technically condescending, but not often or much.

The moderator, however (he is also the program developer, an incredibly productive fellow) is rude and insulting (although the tone of this waxes and wanes a little bit). Forum messages he doesn't like, or that he thinks are not worthwhile, just disappear, sometimes (forum-wise) their authors as well. I am for example one of "los desaparecidos". Still, when I look in the mirror . . .

A chacun son égout!

Best regards,

Doug

Hello again, Doug!

I had similar experiences butting heads with a past Moderator elsewhere. For example, I dared to disagree that DxO was 'the industry standard' for testing stuff. Finally, after a long drawn-out conflict over the dreaded 'inverse square law' he banned me for life.

best regards,

Ted
 

Doug Kerr

Well-known member
Hi, Ted,

I had similar experiences butting heads with a past Moderator elsewhere. For example, I dared to disagree that DxO was 'the industry standard' for testing stuff. Finally, after a long drawn-out conflict over the dreaded 'inverse square law' he banned me for life.
Well, you know that when somebody owns the playing field, and the goalposts . . .

And don't get me started on anomalies on the DxO sensor report page!

Best regards,

Doug
 

nicolas claris

OPF Co-founder/Administrator
This discussion is interesting but should lead to a specific thread, guys you're imho slightly off topic…
What about returning to the original post here?
Doug has always shared here some technicals views and showed is top of the notch technical skills. It quite often was to "techy" for me (the tech itself but the English also) but I appreciate the bonus of such posts.
So back to OT : )
I do not have (once more!) the ability to comment, but what I can say from my experience is that the metering of digital cameras (pros at least) have changed a lot during years…
When I first bought a Nikon D1 in 2000 (or was it in 1999?) the metering did always bring to underexposed image of at least 1 stop…
Now with the Pentax MF, I still have (for bright scenes) to "over expose" to .7 stop…
And I still have a big room to recover highlights.
Doug, could this come from the large sensitivity of the Sony 14 bit sensor?
 

Doug Kerr

Well-known member
Hi, Nicolas,

I do not have (once more!) the ability to comment, but what I can say from my experience is that the metering of digital cameras (pros at least) have changed a lot during years…
When I first bought a Nikon D1 in 2000 (or was it in 1999?) the metering did always bring to underexposed image of at least 1 stop…
Now with the Pentax MF, I still have (for bright scenes) to "over expose" to .7 stop…
And I still have a big room to recover highlights.
Doug, could this come from the large sensitivity of the Sony 14 bit sensor?
I don't think so.

The manufacturer always devises a series of algorithms that are used, based on
"measurements" of scene brightness (perhaps done by the actual "imaging" sensor or perhaps by a separate sensor), to conclude what exposure will be "enacted" for the shot. And of course the behavior of the sensor, certainly well known to the manufacturer, is an important factor ion this reckoning.

This is in a context of trying to attain, over a wide range of situations, a "desirable" exposure result.

But, for one thing, what is a "desirable" result does not have a scientifically-determinable answer. So the manufacturer must, based on many considerations, adopt some objective "target" for exposure results, and then attempt to make their exposure control system attain it in a wide range of situations.

So perhaps what we have here is that the manufacturer of that camera has a different outlook than you do, for certain scene situations, on what is a desirable exposure result.

And I don't mean to at all suggest that your outlook is "not correct".

In this connection, it is interesting to recall that the industry change from one measure of "sensitivity' (the ISO Speed) to another (the ISO SOS) was motivated by a camera manufacturer concluding that the exposure results from exposure control algorithms that were consistent with "traditional" exposure metering norms were "too dark". and decided that the premise of their algorithms should be 1/2 stop "hotter" than suggested by the traditional norms.

But they did this in a crafty way ("inventing" this new measure of sensitivity, the IOS SOS) so that their new, "hotter", exposure algorithms could be shown to still be consistent with the traditional norms.

Best regards,

Doug
 
This discussion is interesting but should lead to a specific thread, guys you're imho slightly off topic…

Sorry about that.

So back to OT :)
When I first bought a Nikon D1 in 2000 (or was it in 1999?) the metering did always bring to underexposed image of at least 1 stop…
Now with the Pentax MF, I still have (for bright scenes) to "over expose" to .7 stop…
And I still have a big room to recover highlights.
Doug, could this come from the large sensitivity of the Sony 14 bit sensor?

Technically, that question bothers me a little, Nicolas.

If the Sony sensor has, say, twice as much "sensitivity" as other sensors, then it's so-called "base ISO" would be 200 ISO by most methods of determining a camera's ISO "rating".

Over-simplifying: that is because it comes down to how much exposure in lux-seconds it takes to saturate the sensor. The more "sensitive" the sensor, the less exposure it takes to do that.

I remember that my Nikon D50 had a minimum of 200 ISO ... implying perhaps twice the sensitivity of most other sensors.
 
Top