Tag EffectiveLV - Values

[Beholder3]

Hi,
I just discovered the tag "EffectiveLV" in the Pentax makernotes which gives a good impression of the lighting situation a picture was taken in.
I also had a look at "AEMeteringSegments".

What I understand is that EffectiveLV already include manual adjustments the user might have made, so an adjustment of -1 and a EffectiveLV of 2.4 means the "real" LV was 3.4.

Now I checked some pictures I took end of last year during indoors xmas parties. Real dark. And the EffectiveLV values do show that. I have a good ISO 12.800 shot at F1.4 that was made at LV 1.4 as it looks - I love the camera. 

Now to what I do not understand:

1. The AEMeteringSegments values sometimes go way below zero (e.g. -2.. How can that be? Actually now I wonder how the zero is defined.

2.  The key question why I am writing here is why another picture at ISO 32.000 and F1.4 can show an EffectiveLV of 63? Doesnt this look like there is something wrong in the conversion?

[Phil Harvey]

The light value may be negative for very dark lighting.  An LV of 0 is defined as f/1.0 at 1 second with ISO 100.  So an LV of -1 would be f/1.0 at 1 second with ISO 200.  (or f/1.0 at 2 seconds with ISO 100, etc)

Pictures taken on a sunny day should have an LV of about 15 or so.

An EffectiveLV of 63 sounds wrong.  What was Exiftool's calculated LightValue for this image?

- Phil

P.S.  I love my K-5 too!

Edit:  I just checked the code.  An EffectiveLV of 63 corresponds roughly to the maximum integer that can be stored by that tag.  Could you send me a sample (philharvey66 at gmail.com)?  It is likely that this should be interpreted as an invalid value.  Also, I think you meant an ISO of 3200 instead of 32000.

[Beholder3]

The light value may be negative for very dark lighting.  An LV of 0 is defined as f/1.0 at 1 second with ISO 100.  So an LV of -1 would be f/1.0 at 1 second with ISO 200.  (or f/1.0 at 2 seconds with ISO 100, etc)

Yes, I found this http://en.wikipedia.org/wiki/Exposure_value#EV_as_an_indicator_of_camera_settings table in the meantime. It seems the use "LV" where they mean "EV".

An EffectiveLV of 63 sounds wrong.  What was Exiftool's calculated LightValue for this image?

I don't understand what you mean by "calculated LightValue"? exiftoolGUI displays "EffectiveLV" 63.0. 

Edit:  I just checked the code.  An EffectiveLV of 63 corresponds roughly to the maximum integer that can be stored by that tag.  Could you send me a sample (philharvey66 at gmail.com)?  It is likely that this should be interpreted as an invalid value.  Also, I think you meant an ISO of 3200 instead of 32000.

That figure is in a 30MB DNG file and it's from a person who I don't have the rights to publish it. Is there any way to either extract only the metadata which would help you or is there a way to actually delete the image contant and just keep the metadata?

And no, I meant ISO 32,000... Pictures taken at an xmas party in a dance club using no flash, just available light handheld using a Sigma 50/1.4. I have some shots I still qualify as "nice and sharp" taken with AF at ISO 12.800 at EffectiveLV 1.4... And yes, I love my K-5, too. 

[Phil Harvey]

I don't understand what you mean by "calculated LightValue"? exiftoolGUI displays "EffectiveLV" 63.0. 

There is a Composite tag called "LightValue" which is derived from Aperture, ShutterSpeed and ISO.  I'm not sure if the GUI shows the Composite tags by default.

That figure is in a 30MB DNG file and it's from a person who I don't have the rights to publish it. Is there any way to either extract only the metadata which would help you or is there a way to actually delete the image contant and just keep the metadata?

There are ways, but I'll see if I can reproduce this myself and get back to you.  You gave me 2 pieces of a puzzle but I failed to put them together:  The raw EffectiveLV value is stored as an unsigned integer, but perhaps it should be signed.  A small negative value would give about 63 if interpreted as an unsigned integer.

And no, I meant ISO 32,000...

Whoa.  OK.  I sort of forgot about the extended ISO feature of the K-5.  12,800 is plenty for me.

- Phil

[Phil Harvey]

I just shot some pictures in the dark with the lens cap on at ISO 51200 and Av wide open, but couldn't get down to negative LV with my DFA100 F2.8.  So I switched to my FA50 F1.4 and could just barely get into the negatives.

Code Select Expand

Effective LV                    : 62.1
Light Value                     : -2.1
F Number                        : 1.4
Shutter Speed                   : 1/60
ISO                             : 51200


Yes, this is the problem:  Even though Pentax writes an unsigned integer format type, it actually stores a signed integer value.

If I change ExifTool to read this as a signed integer (in spite of Pentax's format type), I get this:

Code Select Expand

Effective LV                    : -1.9
Light Value                     : -2.1
F Number                        : 1.4
Shutter Speed                   : 1/60
ISO                             : 51200

Thanks for pointing this out.

- Phil

[Beholder3]

Thanks for your research! 

I still have three things to ask:

1. How do you differentiate between "Pentax writes an unsigned integer" and "actually stores a signed"? In my simple mind the camera outputs some value into the file which is either this or that. What do you mean with "write" vs. "store"?

2. How is the composite tag "LightValue" calculated (I did not find any detailed info here http://www.exiftool.org/TagNames/Composite.html)?
If that is actually the "Exposure Value" as described in the cited wikipedia article then yes you can calculate something, but that would need to be explicitly based on a certain ISO speed (and I guess ISO 100 is quite usual).
If I understand things correctly EV -1.9 at ISO 51200 would translate into -10.9 at ISO 100.

3. How can there be a small 0.2 difference between EffectiveLV -1.9 and Light Value -2.1? Basing it on different ISO speeds should add/substract +1/-1 only for full stops and the minimal change on a K-5 is 1/3, so the minimum difference should also be +0.333/-0.333.

4. If I look at the wikipedia article with your sample figures again, I think 1/60 at f/1.4 would mean EV 7. If calculated back to ISO that gets me a -2. That is close to your "Light Value", but not precisely it.
a) Where does the 0.1 difference come from?
b) This would also point into the direction that the "EffectiveLV" also is already based upon the ISO speed 100, too and needs no further adjustment (while again showing a 0.1 difference, now in the other direction).

[Phil Harvey]

1. How do you differentiate between "Pentax writes an unsigned integer" and "actually stores a signed"? In my simple mind the camera outputs some value into the file which is either this or that. What do you mean with "write" vs. "store"?

When information is stored in a TIFF IFD (as this is), a "format type" code is stored for each value.  This code specifies that the value is an unsigned 16-bit integer for this tag.

2. How is the composite tag "LightValue" calculated

I use the standard formula: LightValue = log2(FNumber^2 / ShutterSpeed / ISO / 100)

If I understand things correctly EV -1.9 at ISO 51200 would translate into -10.9 at ISO 100.

The other way around.  It would be 7.1 EV for the same exposure settings at ISO 100.

3. How can there be a small 0.2 difference between EffectiveLV -1.9 and Light Value -2.1? Basing it on different ISO speeds should add/substract +1/-1 only for full stops and the minimal change on a K-5 is 1/3, so the minimum difference should also be +0.333/-0.333.

Actually, the maxium difference should be 0.333/2 = 0.166, exactly because the camera can only change the exposure settings by 1/3 stop at a time.  For example, if a perfect exposure requires exposure settings of 1.166 EV, the camera is constrained to choose either 1.0 EV or 1.333 EV.

However, this is theoretical only, because I don't know what goes on within the algorithms of the camera's firmware.  Also, the different metering modes apply the readings from the various metering segments differently.

- Phil

Edit:  I found a post in the Pentax forums which may answer some of your questions.

[Beholder3]

Edit:  I found a post in the Pentax forums which may answer some of your questions.

Thanks. That article was quite helpful with careful reading. 

Now I know my AF still works ok until LV 0.9 and creates good pictures and I keep pictures down to -0.4, which lack some sharpness though...

[hpstr]

Hi,

I was happy to read this thread, as I looked for a way to see how much the exposure of a taken picture differs from the exposure meter`s opinion.

However it did not work for me and my K110D (Software 1.02) with a DA AL lens, and two A Series lenses.

As I understand, LightValue (calculated by exiftool?) and EffectiveLV should be pretty much the same in a well exposed picture. In fact they are, even if I see sometimes differences of .1 steps. In a shot with bad exposure, EffectiveLV should be different from LightValue.

In my case however, both values are the same even in a badly underexposed shot. I set my camera with smc DA 18-55mm AL lens to ISO200, Tv, 1/500 and f/3.5, and took a shot in a quite dark angle of my room, producing an apparently black picture.

However, I get identical values, resembling the nominal EV value for the chosen settings:

Code Select Expand


hps@tango:/tmp/xxx$ exiftool -LightValue -EffectiveLV IMGP0379.JPG
Light Value                     : 11.6
Effective LV                    : 11.6


I have seen in the exiftool history that EffectiveLV once was called MeasuredEV, so I suppose that the tag is populated by the camera, and not calculated by exiftool.

I upgraded to 8.99 today with the same results.

Could someone please try to repeat this procedure (forcing an underexposed shot) to confirm this, preferrably with a K110D or a K100D?

Any ideas?

Thank you, regards,

Hans

[Phil Harvey]

Hi Hans,

In the Pentax tag name documentation for EffectiveLV, it states: "camera-calculated light value, but includes exposure compensation"

Does this help clear things up?

- Phil

[hpstr]

Hello Phil,

thanks for coming back to my question.

"camera-calculated light value, but includes exposure compensation"

Does this help clear things up?

To be honest: not really. I have seen the tag list yesterday, and thought that "exposure compensation" would refer to the manual -/+ 2 EV compensation, using the Av button. Mine is set to 0.

If EffectiveLV considers any manual user intervention (e.g.: setting a shutter speed priority) as "exposure compensation", then in any mode other than the Auto modes EffectiveLV and Light Value should never be different. And in Auto mode, I would not really know how to produce a badly exposed shot (if not abusing the lens cap).

So if I get it right, E.LV and LV will -practically- never show a significant difference? If they will in certain circumstances, then I didn`t...

To help me understanding the tag, could you please give me an example where they might be different? I have tried about 25 shots tonight in several configurations, each having equal EffectiveLV and LightValue (not considering the occasional 0.1 EV step differences mentioned above).

Thank you, best regards,

Hans

[Phil Harvey]

Hi Hans,

If EffectiveLV considers any manual user intervention (e.g.: setting a shutter speed priority) as "exposure compensation", then in any mode other than the Auto modes EffectiveLV and Light Value should never be different. And in Auto mode, I would not really know how to produce a badly exposed shot.

If you aren't using exposure compensation then the only other way to purposely get an under or over-exposed image is to shoot manual (set both shutter speed and aperture).  But with my K-5 in manual mode, EffectiveLV just reflects the manual settings.  So the only time EffectiveLV is different from LightValue is when I use exposure compensation.

- Phil

[hpstr]

Hi Phil,

ok, so I got it right. Too bad, together with your LightValue EffectiveLV could be very useful, if the Pentax folks would have used the light meter`s reading instead of the camera settings. Having a dedicated tag for exposure compensation as well, EffectiveLV seems somewhat useless to me - at least for us exiftool users (I don`t know if other metadata tools calculate EV).

Thank you for helping me understanding the tag - and, of course, for exiftool itself! I am not aware of any other one-man-written tool with so many releases and help (as reading the forum shows) by the author himself.

Best regards,

Hans

[Beholder3]

But with my K-5 in manual mode, EffectiveLV just reflects the manual settings.  So the only time EffectiveLV is different from LightValue is when I use exposure compensation.

Huh? I can definitely say that EffectiveLV contains metered values (corrected by Exposure Compensation) and it does differ from LightValue for my pictures 99% of the time.
I use exiftoolGUI and have both LightValue and EffectiveLV as columns, so I can see both values for all my pictures quite easily in an overview.
The differences usually are less than 1 EV, but they are there (the typical differences between photos).

And if I intentionally set wrong paramters in manual mode I get really different values, not using exposure compensation at all.

ok, so I got it right. Too bad, together with your LightValue EffectiveLV could be very useful, if the Pentax folks would have used the light meter`s reading instead of the camera settings. Having a dedicated tag for exposure compensation as well, EffectiveLV seems somewhat useless to me - at least for us exiftool users (I don`t know if other metadata tools calculate EV).

As I just wrote above: It does contain the light meters readings, actually the averaged out results of the segments.

The tag "AEMeteringSegments" contains the meter reading directly, nonadjusted for each of the 77 segments.

I once created a simple excel spreadsheet which roughly shows the brightnesses per segment after pasting in the values from exiftool.

It is all there. 

[Phil Harvey]

And if I intentionally set wrong paramters in manual mode I get really different values, not using exposure compensation at all.

Thanks for picking up on this.  It must be that the images I checked just happened to be exposed properly.  (I didn't run any specific tests.)

I once created a simple excel spreadsheet which roughly shows the brightnesses per segment after pasting in the values from exiftool.

As you can tell, my memory isn't all that good.

- Phil

[dosdan]

Edit:  I found a post in the Pentax forums which may answer some of your questions.

I wrote the LV calculator referred to in that link. 

An interesting related topic to LV is the old APEX system. I've explained it here:
http://www.pentaxforums.com/forums/photography-articles/126203-stops-shutter-speed-aperture-iso-sensitivity.html#post1717721

Exif 2.3 already contains APEX data which can be used, with a bit of mental arithmetic, to work out LV:
http://www.pentaxforums.com/forums/photography-articles/126203-stops-shutter-speed-aperture-iso-sensitivity.html#post1721119

Dan.