exiftool incorrect Light Value

[WayneF]

Exiftool LV calculation is incorrect, it uses wrong sign for ISO.  (LV is last value in ExifTool Composite).

Definitions:

EV = log2 (fstop² / shutter time), which is independent of ISO.

LV = EV + log2 (ISO / 100).   


But Exiftool has the ISO sign wrong (it must use 100/ISO, which log of a fraction is negative, which subtracts instead of adding).

Example:   f/4 1/50 second ISO 1600

f/4 1/50 second computes EV 9.6  (EV charts, independent of ISO)

LV for ISO 1600 adds +4 getting LV 13.6, which absolutely agrees with Sekonic meters.

But ExifTool must use 100/1600, because it uses -4, getting LV 5.6, which is incorrect.   Big difference.

FWIW, http://www.fredparker.com/ultexp1.htm#EXPOSURE%20FACTOR%20RELATIONSHIP%20CHART%20B  is wrong too, same wrong ISO sign.

Greater ISO increases the light value, necessarily requiring less exposure (faster shutter speed, etc). which is a greater EV number.

My own interest is my calculator at http://www.scantips.com/lights/exposurecalc.html

People tell me it must be wrong because Exiftool gets a different value.     But it's not wrong, and Sekonic and Wikipedia instead agree with the correct math.

FWIW, Exposure Compensation is Not a factor as such.  It does change camera EV settings, which is then taken into account automatically.

[WayneF]

Discussing it with myself.   Reconsidering that this should not be a "bug report", instead it was possibly an intended effect.  I do love Exiftool, obviously best one, and actively supported and able to keep up with cameras, thank you, it's great, only good choice out there.  But maybe more likely Exiftool is just doing the unexpected with LV -  at least I would not expect it.   Admitted, this LV is a very tiny thing, nothing compared to the rest of what Exiftool does so well.
   
EV= log2(f^2/T) already has ISO applied to those camera settings, certainly if about settings that the camera used or should use, or if about what light meters read as EV at an ISO (that the camera should use).   Any EV actually used necessarily already has ISO factored into it, to be the correct settings for that situation and ISO. That seems like the EV value that should be shown... what the camera actually did do. What a meter would read to match it. The actual useful number.

But then we also have LV = EV + log2(ISO/100) which shows effect of ISO (if different than ISO 100, to be relative to ISO 100). But ISO has already affected the EV settings.  This formula seems just a trouble maker (because it adds ISO twice to EV), worse than helpful. All it does is convert an ISO 100 EV to be  EV at an ISO, but which any EX used already had factored in.

But Exiftool is subtracting ISO from LightValue, possibly intentionally subtracting, to back ISO out of EV, to show an  ISO 100 EV value? Why, I dunno? Not meaningful if this camera exposure did not use ISO 100.  It is only correct if at ISO 100.

The LV formula seems only useful for converting ISO 100 to a different ISO.   Any EV actually used already has ISO factored in anyway.   I previously assumed Exiftool was incorrectly subtracting ISO effect from LightValue.  But possibly it intended to subtract ISO, to show EV at ISO 100?  Why ISO 100, I dunno.  Seems wrong either way.

It seems more useful to show the EV that the camera actually did do, and to agree with the EV that light meters would show at the ISO. 
This is just the EV= log2(f^2/T) number, it already factors in the ISO used.   Agrees with the EV chart. Agrees with light meters.  The ISO 100 EV conversion seems a useless number (unless it really was using ISO 100).

[Phil Harvey]

Hi think we disagree about the way to calculate LV.

The Wikipedia page for "Exposure value" gives the formula for calculating LV (or EV100), which is the same as the formula that ExifTool uses.

LV is the light falling on the camera.  Changing the ISO doesn't change this.  If you leave aperture and shutter speed the same, but need to increase ISO to get a proper exposure, then the amount of light has decreased, so the LV goes down, not up.

- Phil

[WayneF]

Yes, big disagreement, which is my reported bug.  And I'm sorry to be a pain, but you are wrong, only about this (and yes, EV ISO conversion is certainly most commonly very wrong on the internet).  However, that Wikipedia Exposure Value page is correct, it just does not say what you report.  It clearly says ISO INCREASES exposure values (decreases the Exposures).  See my attachment here, from it.  Clearly says this, says "for example,  ISO 400 is two stops greater than ISO 100". And of course, it is.  And of course, Sekonic meters will read EV as two EV greater at ISO 400 than at ISO 100.  The EV chart shows the +2 EV exposure two rows lower (greater EV number).

My own interest is that  I have a much better calculator at http://www.scantips.com/lights/exposurecalc.html . It also at first made the same naive error of trying to add any ISO effect a second time, to camera Exif values that already included the ISO effect. That produced garbage numbers too.  But it is corrected now.

But Exiftool, as great as it is, does get this ISO LV point wrong.  I can appreciate you don't want to be bothered with this minor thing, it's nothing in the big picture of Exif.  But it is wrong, LV shows very incorrect numbers.  I'm sorry to be a pain, but it is trivial to fix it, or even simply deleting wrong LV is far better than no action. But LV as shown is total nonsense, useless for any purpose.  It should show EV (and just skip LV). EV is useful.

I hope you listen, the details are here.

Exposure Value is not even about "the light".  EV is about the camera EXPOSURE, specifically reporting about the camera SETTING NUMBERS to provide exposure to be correct for that light. Frankly, camera EV gets the same numbers if the lens cap is on. The camera Settings do reduce the exposure for greater light. So of course EV increases with ISO (less exposure required). But the camera Exif already shows settings for the actual ISO.  No adjustment is necessary, or desired. Any tampering screws up a good thing, corrupts the right numbers. It is simple if we just stop and think.

The LV formula is NOT what you want.  It is always written as merely an ISO conversion "FROM ISO 100".  But if the Exif in camera was NOT at ISO 100, the camera settings already factor in ISO conversion.  Adding any ISO effect a second time gives garbage numbers.  Adding a ISO 1250 conversion "from ISO 100" to a camera value already at ISO 1250 is simply nonsense, regardless if adding or subtracting.  Either way is nonsense, either way then DOES NOT agree with Sekonic meters, or the EV chart (but of course, it must agree).  So for existing Exif settings, already what the camera actually did do, any ISO conversion is wrong because, the camera SETTINGS already figured in that ISO.  The EV number (not LV)... the EV number is the settings that the camera actually did use for whatever ISO was used.

Any EV settings should reflect what the camera actually did do, and should agree with a Sekonic meter, and should agree with the EV chart.  However, Exiftool LV does not agree. Throw out the LV term, and show EV, and it will agree, and will be perfect.  Look up the camera settings in the EV chart, and then the chart will specify the correct EV value.  (regardless of ISO used).  Any other value is wrong. Exiftool is wrong about LV.

LV adds any greater ISO effect, it does not subtract it (Exiftool does?)   But the Exif is about the camera settings, which already handles the settings for the light (i.e., actual Exif settings already include effect of ISO used). EV will report that correctly.  And the LV formula (as written) is merely a converter specifically from ISO 100. A general converter is log2(ISO2/ISO1), and NOT  ISO/100.  It is a conversion FROM 100, NOT TO 100.  But 1) if the camera Exif was not at ISO 100, then the LV formula as written is useless, and wrong.  Because 2), the camera exif settings already dealt with the ISO used. EV is correct as is.

The proof is in the simple obvious results.  If you have a Sekonic meter, meter in bright sun at ISO 100, and it will report near EV 15.  The chart (and Sekonic) says this is f/16 at 1/125, and that is a correct equivalent exposure (near Sunny 16).  Change ISO to 800, and meter again.  It will report near EV 18, and for EV 18, the EV chart says f/16 at 1/1000, and again, it is correct exposure (greater EV).  This is how it works, and is what the camera will do, and what the Exif says the camera did do.   

But Exiftool will somehow claim that camera Exif at f/16 1/125 ISO 800 is LV 12 ? ,  That is a totally wrong and useless number.  Look up EV 12 in EV chart, and it says f/16 at 1/15 second, suitable maybe for deep shade. But our camera did NOT even use 1/15 second, and was in bright sun, and is NOT at EV 12. Bright sun did get the correct EV18 exposure at ISO 800. So Exiftool reports an extreme nonsense number for lightvalue. It should simply report EV instead, from formula or from chart. The LV 12 has less than zero use, detrimental to any purpose.

Exiftool should simply discard the garbage LV term, and report EV.  Then the Exif camera settings will report the correct EV that matches Sekonic and the EV chart, for that actual exposure the camera did use.  This seems a huge improvement.  Or simply just deleting the incorrect LV would be a big improvement too.

Thanks for all the effort on ExifTool.  It's the only one useful for Makers Notes in late model cameras (the other Exif viewers are years out of date). I am sorry to be a pain.  But ExifTool does have this one tiny little problem about not understanding LV.

[Phil Harvey]

Interesting.  All of the references I can find agree with ExifTool's calculation.  Even the lviso.gif screen shot you posted agrees with ExifTool (LV == EV100, so the ISO is subtracted when calculating LV).

It seems that it is you against the internet here.  For example, Ken Rockwell says this (http://www.kenrockwell.com/tech/ev.htm):

An LV, or Light Value, is a number that represents how bright a subject appears in absolute terms. It does not take film speeds or exposure into account.  LVs are very handy photographic terms to use to describe lighting levels.

The lighting level doesn't change when you change ISO.

- Phil

Edit:  I took a look at your description on scantips.com.  The Sekonic meter shows EV (as it says), not LV.  LV is the same as EV at ISO 100.  So set the meter to ISO 100 and you'll effectively get LV.  Change the ISO and EV changes, but LV does not.  You say that this is a useless number, but on the contrary, it is very useful because it gives an absolute measure of the light level.

[WayneF]

Interesting.  All of the references I can find agree with ExifTool's calculation.  Even the lviso.gif screen shot you posted agrees with ExifTool (LV == EV100, so the ISO is subtracted when calculating LV).

Slow down, it does not say any of that.  It very clearly says greater ISO increases EV, which it of course does.  It never mentions subtract, the formula says +.  And log of 800/100 is positive (log of a fraction like 32/100 is negative).  It does show how LV converts ISO 100 values to other ISO, however if our Exif settings are already ISO 800, then do you really want to convert a ISO 800 value to 800 AGAIN, pretending it was only ISO 100 first? (you really don't).    We must stay away from screwing it up with wrong numbers.

Here's the deal.    If you choose to defend ExifTool reporting EV 12 for Exif camera settings of bright sun at ISO 800, you should be able to describe how EV 12 is used, of what possible use is it?  What does it mean?  What is one possible thing we can do with the EV 12?   That should be very interesting, however the only answer possible is it is of absolutely no use, it is totally a wrong number in every context.  I am trying to advise you that it should be embarrassing.

However, if ExifTool instead correctly chooses to report bright sun at ISO 800 as EV 18, which Sekonic also meters, then it would perfect.
To do that, simply report the first EV formula of the settings actually used in Exif, and totally ignore the LV formula, which is inapplicable here.

I am trying to get your attention, for you to think about that.  It is very a easy concept, think of it in the easiest possible way.
EV 18 is what Sekonic meters will report (in bright sun at ISO 800).  And EV 18 is the row where the EV chart reports useful exposure settings to use for EV 18.  And of course it works great. 
However Exiftool reports EV 12 there.  Please think about of what use is reporting EV 12 in bright sun at ISO 800? It is only a wrong number you made up. 
At ISO 800 in bright sun, the Exif has camera settings for EV 18. Where did that correct number go? What is this 12 business?

Bright sun at ISO 800, easy for you to try it.

It seems that it is you against the internet here.  For example, Ken Rockwell says this (http://www.kenrockwell.com/tech/ev.htm):

That is sad isn't it?    Yes indeed, sad because so few understand a simple concept of EV.    All this wrong crap on the internet does not help.

An LV, or Light Value, is a number that represents how bright a subject appears in absolute terms. It does not take film speeds or exposure into account.  LVs are very handy photographic terms to use to describe lighting levels.

Yes, not exactly wrong, but misleading. It fails to say the right words, those which would make it useful.

Even the Wikipedia article you mentioned says of the EV chart:

"EV corresponds simply to a combination of a shutter speed and an aperture setting, independent of any ISO setting".

Which is very correct, but also misleading, in that it just means, it is for any ISO, for whatever ISO is. The EV is the one chart row where our f/stop and shutter speed appear, regardless of ISO.... but specifically for that ISO in use.
For whatever ISO we use, if we meter EV xx, then EV xx is the one chart row where we must choose the exposure settings, for this ISO.

It absolutely does NOT MEAN that EV does not change with ISO. Obviously EV does change with ISO.  That is how we use it. Sekonic meters it. ISO can vary it.   I guess we could think of EV as the light behind the camera lens.   Then we do use the EV chart for any ISO value we are using. And it works. This is how it is used.

The Sekonic meter (bright sun, ISO 800) meters EV 18.   This is exactly LV as intended to use it, an ISO 100 value increased 3 EV for ISO 800, It is exactly EV100 + log2(800/100).  But NOT useful (is detrimental) if the Exif already has accounted for ISO 800 settings. LV would add ISO 800 effect twice then, grossly wrong concept.


Anyway, the Sekonic meter meters EV 18 (but which Exiftool will report as LV 12??? )  Bright sun at ISO 800, easy for you to try it.   You ought to instantly recognize your error if you stop to think then.  Because then the correct exposure is any equivalent exposure value on the EV 18 row of the EV chart.  It says f/16 at 1/1000 there, which is a correct exposure for ISO 800 in bright sun.  It works. But Exiftool will strangely report LV 12 there, which is nothing about nothing, and should be corrected.  It should report EV 18, because it is.  The settings already account for the ISO 800, and the value is EV 18.  Or Exiftool should at least simply remove the wrong LV crap.  EV 12 is a pretty funny thought for bright sun at ISO 800.  EV 12 says f/16 at 1/15 second.  At ISO 800 in bright sun? Totally meaningless and useless.

The lighting level doesn't change when you change ISO. 

- Phil

Right, but of course EV is NOT about the light.  EV is Exposure Value, and is about the Exposure, specifically about the camera settings, only about fstops and shutter speeds,  for settings of an exposure useful for the light. When light shifts one way, our exposure shifts the opposite way.  But its about the camera settings, which are the same even if we put the lens cap on.  EV formula is about the camera settings.  The settings are in the Exif, and Exiftool tries to report EV equivalent, but gets it very wrong.

Edit:  I took a look at your description on scantips.com.  The Sekonic meter shows EV (as it says), not LV.  LV is the same as EV at ISO 100.  So set the meter to ISO 100 and you'll effectively get LV.  Change the ISO and EV changes, but LV does not.  You say that this is a useless number, but on the contrary, it is very useful because it gives an absolute measure of the light level.

Maybe you looked, but apparently you did not see.  In bright sun, Sekonic meters EV 15 at ISO 100, and EV 18 at ISO 800. 
This is exactly the LV = EV100 + log2(800/100) formula.  This is where it is used.   The meter is not already at ISO 800 values, we merely told it to convert to ISO 800, so it does. 
We look up either of the EV in the EV chart to determine the proper settings to properly expose these.  It works. This is how it is used.

But the Exif settings are already at ISO 800, so we don't do it again.

This LV formula was used there, but it CANNOT be used again if our Exif setting values have already accounted for ISO 800.  That adds +3 EV twice (to show EV 21), converting FROM 100, not TO 100.   (except it was already 800, and  you choose to subtract it to get the totally irrelevant and meaningless and useless EV 12 number).    Very wrong numbers, adding or subtracting.  The setting are already for ISO 800, and already EV 18.

I hope this is sill friendly. But you are resistant, so I am playing Trump here, telling it like it is, not politically correct. 

I am absolutely huge fan of ExifTool but think it should be perfect (see http://www.scantips.com/lights/exif.html )

You really do need to give this some thought.  Please?  The ExifTool LightValue is grossly incorrect.

[Hayo Baan]

I haven't looked at the figures that exiftool gives me, but to be honest it seems like you are mixing up light value and exposure value. To me it sounds very logical that light value is indeed the absolute measurement of the light, and ev is the measurement of light, taking into account the ISO sensitivety (and thus can be used to calculate shutterspeed and aperture).

In other words, I would expect LV to be the same for two shots taken at the same scene, but at different ISOs, and I would expect the EV to change (higher for the higher ISO).

This is what I would expect exiftool to report. Is this not what it does?

[WayneF]

I haven't looked at the figures that exiftool gives me, but to be honest it seems like you are mixing up light value and exposure value. To me it sounds very logical that light value is indeed the absolute measurement of the light, and ev is the measurement of light, taking into account the ISO sensitivety (and thus can be used to calculate shutterspeed and aperture).

In other words, I would expect LV to be the same for two shots taken at the same scene, but at different ISOs, and I would expect the EV to change (higher for the higher ISO).

This is what I would expect exiftool to report. Is this not what it does?

That is not what Exiftool does. Exiftool has absolutely no clue about the light.  It sees no light.  It only sees camera settings in the Exif.  This is NOT exactly semantics.   The camera settings can be the same even if the lens cap is on.

Exiftool reports LightValue EV 12 for a correctly exposed picture in bright sun at ISO 800.  There is no possible way to explain how that incorrect value 12 is useful or has any meaning. Bright sun at ISO 800 is about EV 18. That is a difference of 6 stops, near infinite. 

It is not at all complicated or difficult, but it does require slight thought.

Light value has a standard formula and purpose, like discussed at http://en.wikipedia.org/wiki/Exposure_value  (down below the EV charts)

EV = log2(f^2/T)
LV100 = EV + log2(ISO/100)

So what this LV is - is just a conversion of EV for ISO, specifically FROM ISO 100 to a new value. That is what Sekonic meters do when metering EV for ISO 800. It is a conversion FROM ISO 100.   It is NOT a conversion TO ISO 100 as imagined.  Look at the formula.

But EV is NOT EV100 if the camera used an ISO not 100.  LV doesn't care, it is happy to add ISO effect again, twice if we do it again.

There is not really anything called Light Value as such.  The meter does meter the light, but the EV result is only about camera settings to capture the light.  About the settings, not about the light.  Changing fstop or shutter changes EV and exposure and LV. (changing among equivalent exposures does not).  Changing ISO changes LV and exposure.  What is called LV is the EV (the settings) adjusted for ISO different than 100.
Light meters creating new settings need to do that.  Existing exposures already in the Exif do not, ISO has already been done once.

Specifically,
Sekonic, metering bright sun at ISO 100 is EV 15.  Metering same bright sun at ISO 800 is EV 18. 
Looking up EV 15 is f/16 at 1/125 sec.  Looking up EV 18 is f/16 at 1/1000 sec. Both are correct exposure in bright sun. That is the point.

Exiftool describes the settings for that ISO 800 picture as LV 12. Grossly wrong.  EV 12 is f/16 at 1/15 second, totally ridiculous for bright sun at ISO 800. which it was.   So it was not EV12, nor was it 1/15 second nor was it ISO 100.   My question would be for some way to explain the meaning of this EV 12 reported?  But it is just an error, doing the math wrong.  No way to explain it without seeing the incorrect math.

It is wrong to apply (or use) this LV formula to existing camera or metered values, because they already factor in ISO.     The settings are chosen to be correct for using the ISO, among other things.  But doing LV factors the ISO in AGAIN (wrong).

So metering bright sun at ISO 800 is about EV 18.  The simplest try of this confirms that.  But then applying the LV formula again factors ISO in AGAIN, getting EV 21, which is wrong. So apparently users say Oops, maybe it should be subtracted, and they get EV 12.  Maybe it is ISO 100 or something?  But of course, it's not (ISO 100 was EV 15).  It is thee stops under ISO 100, or ISO 12.  EV 12 is just a math blunder. 

Exiftool shows EV 12 (for exposures in bright sun at ISO 800).  It is not just that one value 12, the method is wrong, every value is wrong (except for ISO 100 where LV adds zero).  However there seems to be a few of values that vary slightly. Puzzles me.

ExifTool is reporting a bogus LV number.  All Exiftool should do is to report the simple EV number,  EV = log2(f^2/T)  which describes the exposure, the useful number which is limited to one row of the EV chart, which is a row of Equivalent Exposures, all the same exposure for the EV.  Any combination of fstop and shutter speed exists only on one row of the chart. And that row EV will match the ISO of those settings. That is how it is used.

For camera settings in the Exif, these camera settings already have ISO factored into it. Nothing else need be done.  Just show EV.

Or, omitting LV and doing nothing would be an improvement too. But showing EV would be a nice touch.

This seems so very clearly obvious.  ExifTool is reporting LV wrong.   I'd like to see it corrected.  Or I'd like to hear how ExifTool LV 12 has any meaning and use for the bright sun ISO 800 picture...

[Phil Harvey]

Slow down, it does not say any of that.  It very clearly says greater ISO increases EV, which it of course does.

Sorry, but that isn't what I said.  I said that ISO should be subtracted in the formula to calculate LV.  Please pay attention.  You are talking about EV again, and I'm talking about LV.

- Phil

[WayneF]

Slow down, it does not say any of that.  It very clearly says greater ISO increases EV, which it of course does.

Sorry, but that isn't what I said.  I said that ISO should be subtracted in the formula to calculate LV.  Please pay attention.  You are talking about EV again, and I'm talking about LV.

- Phil

? How so? How do you figure that? Could you explain in any degree to clarify why? What does the LV number Exiftool computes mean? What possible use does it have?  What can we do with the number?  Please say some words to explain yourself.  How does that work? You probably can explain any other number you show. What does this LV number mean, specifically?  How would it be used?  I tend to be detail oriented, help me out? What do you mean?

Let's say it another way?  What is the formula that ExifTool uses to show what it calls Light Value?   Or is this a secret method of some newly devised concept? 

Showing EV = log2 (f^2/T) would clearly be very meaningful, since this value is what a meter would show if metering that scene at the ISO used. It is indeed pretty good info.  But what does ExifTool Lightvalue mean?

Sorry, I don't see how "subtracted" is mentioned there in any way?  Other than that you say it does. I don't see it.

Of course, the formula does clearly say "plus".



See the EV100 there? See the ratio ISO/100?  See the +?  All the words there say "increases". It is used to convert ISO 100 values to other ISO.
But the Exif data is already at another ISO, specifically the ISO of interest.  Far better to leave well enough alone.

So the formula clearly intends a different use, specifically only to convert an ISO 100 value to other ISO.
But the Exif data already has used ISO, like ISO 800.  The ExifTool number only comes out right if the camera had used ISO 100 (log of 1 adds zero).
This formula arbitrarily converts from ISO 100 values (the number 100 is arbitrary).  It could have used any number, but 100 is convenient. However this is the only reason we imagine the EV chart is ISO 100, because its numbers agree with this arbitrary conversion at 100. You can call EV100 to be "light value", whatever that means, but this formula is just a simple conversion, FROM 100, NOT TO 100.   But the EV chart is obviously used with ANY AND EVERY ISO value.

But the Exif setting values are already at the final ISO.  It is the ISO we are interested in.  Converting it as it were 100 is simply wrong.  That is only correct if it was 100. Otherwise, it is garbage.

(my plan is that even fast skimming might notice if repeated enough.  But careful thought will be the most productive).


Light value only has the meaning of the previous chart there, where it shows typical EV values like:   Night Sports    EV100  9
But I suspect metering will be more useful.

You first said all the references you find agree with ExifTool. That might sadly be true, but please name one.  I'd like to see it, so please be more specific.  I assume you meant Fred Parkers EV site, it comes up near the top of Google, and which does get the sign wrong in every case. 

As a reference, it is indisputable that bright sun is universally said to be about EV 15 at ISO 100 (Sunny 16 is only 1/3 stop different, from before ASA, from back when Weston meters defined their own standards).   

Here is obvious proof Parker has all the signs backwards.

Parkers EV chart says 15 EV at ISO 100 maps to EV 12 at ISO 800.  Obviously wrong, obviously backwards, obviously useless, and certainly that would greatly surprise Sekonic (which meters EV 18).   
Look up EV chart exposures for EV 12 (f/16 1/15 second). Does that sound right for bright sun at ISO 800? 
Instead correctly add it, to get EV 18.  That says f/16 at 1/1000 sec, which of course is correct exposure for bright sun at ISO 800 (and it agrees with Sekonic and others).
Adding is how it works (we do subtract ISO values less than 100, but log of a fraction is negative, so this works out automatically when the formula adds). 
Subtracting 3EV for ISO 800 give a result SIX stops wrong.  Adding the 3EV comes out perfect.

Except that formula is for establishing initial EV for a new ISO (like a light meter does). The Exif settings have already done that, they already have setting values for the ISO 800 used.  Converting that number to ISO 800 a second time is a stupid bad plan. That's the only reason adding doesn't work for you.  But subtracting is also a plan just as bad.  Exif data already has ISO in its values. Once is enough.  Using this formula is the problem.

Parkers page also has the chart of "Brightly lit home interiors at night. Fairs, amusement parks.", EV 6.    He just does not bother to mention the data is for ISO 100.  Pretty crude and unknowing.  The page is about 20 years old, and he has never bothered to fix it. Sadly, it seems to have influenced too many others who didn't understand either.  So maybe it is just me and Sekonc, but any practical results definitely also agree with us.

Too chatty, sorry.  But the values in the Exif for say ISO 800 clearly are not about ISO 100, so this formula to convert from ISO 100 is simply the wrong plan.  It is not light value, It is a conversion, but Exif data already uses the final ISO.  How hard can this be?

My notion is that the ExifTool value obviously ought to agree with Sekonic meters, which will also agree with the meter that the camera used.  I think it is a neat touch that you show EV there, but if it is going to show EV, it should be a correct and useful and meaningful number... the EV that the camera actually did use.  Not some meaningless number.

What is the use of the number ExifTool shows now?  (rhetorical, I know no answer is possible).

It seems extremely clear, but of course, if you have no desire to bother to fix it, then we might as well stop now.

[Phil Harvey]

Of course, the formula does clearly say "plus".

You're not paying attention.  LV == EV100, which is what we are calculating.  So to calculate LV you need to rearrange the formula to this:

LV = EV100 = EVs - log(S/100)

which is the formula the ExifTool uses.

The use is clear.  Given a light value, you can set your camera to the correct exposure.  I don't have much time right now to explain how, but I'm surprised that you need an explanation of this.

- Phil

[WayneF]

Of course, the formula does clearly say "plus".

You're not paying attention.  LV == EV100, which is what we are calculating.  So to calculate LV you need to rearrange the formula to this:

LV = EV100 = EVs - log(S/100)

which is the formula the ExifTool uses.

The use is clear.  Given a light value, you can set your camera to the correct exposure.  I don't have much time right now to explain how, but I'm surprised that you need an explanation of this.

- Phil

Then at bare minimum, to have any meaning, you should at least reword "LightValue"  to instead say:     EV at ISO 100.
If that is your meaning, I think anyone would see the value of changing that.  It's only a few more characters, not at all the longest one.

Light Value has been used that way, in old analog meters (with dials), but who knows it now? A few old Sekonic digital meters said Light Value, but they showed it at working ISO (to be useful).  There is certainly no clear definition of it.  Camera meters have no use for it.
Any users would surely better understand   "EV at ISO 100"  if that's what it is.

Why 100, I dunno.  It is not a standard reference unless you point out you mean it that way. Then still dunno why?  Looking up EV for ISO 100 in the EV chart does not give the same camera settings that the Exif data used, etc.  (unless the camera did use ISO 100).  If using say ISO 2400, we have no use for a ISO 100 number.  But when we are in the Exif, we do know the camera did use ISO 2400.  It seems the only number that matters then.

OK, so be  it. My surprise was that you are uninterested in it. 

[Hayo Baan]

Hi Wayne,

I've been following this discussion with some interest.

Then at bare minimum, to have any meaning, you should at least reword "LightValue"  to instead say:     EV at ISO 100.
If that is your meaning, I think anyone would see the value of changing that.  It's only a few more characters, not at all the longest one.

Interesting you mention this, I disagree though. Without studying EV vs LV extensively (if at all), it was quite clear to me what the meaning of LV was and I actually think rewording it to EV at ISO 100 might even confuse things...

Light Value has been used that way, in old analog meters (with dials), but who knows it now? A few old Sekonic digital meters said Light Value, but they showed it at working ISO (to be useful).  There is certainly no clear definition of it.  Camera meters have no use for it.
Any users would surely better understand   "EV at ISO 100"  if that's what it is.

Why 100, I dunno.  It is not a standard reference unless you point out you mean it that way. Then still dunno why?  Looking up EV for ISO 100 in the EV chart does not give the same camera settings that the Exif data used, etc.  (unless the camera did use ISO 100).  If using say ISO 2400, we have no use for a ISO 100 number.  But when we are in the Exif, we do know the camera did use ISO 2400.  It seems the only number that matters then.

Because LV doesn't change with the ISO used, to me it makes it a much better tool for comparing images, don't you think? Much better than EV because that would be all over the place, especially when shooting with Auto ISO...

OK, so be  it. My surprise was that you are uninterested in it. 

I don't think Phil is uninterested at all (otherwise he wouldn't have taken the time to answer you here), he just disagrees with you. And I must say, I do agree with Phil on this one. Not only because he correctly interprets the formulas (as you now admit yourself), but also because I think LV is a much more valuable value to know than EV

Cheers,
Hayo

[WayneF]

Because LV doesn't change with the ISO used, to me it makes it a much better tool for comparing images, don't you think? Much better than EV because that would be all over the place, especially when shooting with Auto ISO...

Except the Exif and ExifTool sees no light.  All values can only be about the camera settings.  The formulas compute only camera setting numbers.  This light value is merely a hopeful presumption of an expected proper exposure.  We would of course compute the same numbers even if we put the lens cap on (assuming a manual exposure doesn't change settings).  Or if we change the settings, and then the presumption is also forced to change, but not due any property of the light.  Light-wise, we are sort of in the dark.

Possibly semantics, but I am not very sure we know any light level.  I do agree that the ISO 100 value is hopefully a possible representation of the light level, in that it specifically hides any ISO boost and related exposure settings that may be needed for proper exposure of the scene. It should work if ISO 100 is possible, it is just EV then. But instead, the EV number tells us what the camera actually did do, how much EV was needed at the ISO we chose to use. We can already see this in the Exif, but the specific beauty is that the EV number is a usable number, it matches light meters, matches the EV chart, and tells the story, etc. When using a light meter, it does not tell me the light, it tells me the EV settings needed for exposure at the specified ISO.  That's why we use light meters.   A photographer is probably more concerned with exposure (the settings needed), and interested in the ISO boost and settings and EV that was needed for the picture. There does seem to be a good case for EV.

Of course, both EV and LV will need an ISO setting, but calling it EV at ISO 100 would at least describe what this item actually is.  Knowing that, I could at least then know how to adjust the settings to my working ISO used.  Of course, the settings and ISO are known in the Exif, so I can ignore it and compute EV myself, but ExifTool had this idea, and it seems an opportunity to help, if we knew what it was. I guess showing both values is out, that would require more explanation, which is usually the weak point. 

Yes, it is pretty minor in the big picture. Thanks for being the diplomat. 

[Phil Harvey]

I'm glad we're at least on the same page now.  I agree that this doesn't translate into actual light levels unless the image is properly exposed, and for my own purposes I adjust LV by the ExposureCompensation to take this into account when comparing light levels of different images.

- Phil