Pages: [1]   Go Down

Author Topic: LR 4.4 Incorrect color temperature reporting  (Read 5621 times)

André Denis

  • Newbie
  • *
  • Offline Offline
  • Posts: 17
LR 4.4 Incorrect color temperature reporting
« on: August 04, 2013, 09:14:45 am »

Lets say I set the Nikon D800 WB to manual K=5560.
I import the pictures with Photo Mechanic.  It reports the picture  'as shot' WB as 5560K.  This is fine.
I import the very same picture in LR 4.4, and it reports the 'as shot' WB as 5350K with Tint=+6!.
Note that this info comes from the Develop Tab view.   In the Library Metadata Tab, there is no mention of the 'as taken' white balance!
This is a Raw file, not jpg.

Can someone please explain this discrepancy between LR and Camera 'as shot' white balance info?  Am i doing something wrong?

Thanks and regards.
 
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: LR 4.4 Incorrect color temperature reporting
« Reply #1 on: August 04, 2013, 11:23:37 am »

Lets say I set the Nikon D800 WB to manual K=5560.
I import the pictures with Photo Mechanic.  It reports the picture  'as shot' WB as 5560K.  This is fine.
I import the very same picture in LR 4.4, and it reports the 'as shot' WB as 5350K with Tint=+6!.
Note that this info comes from the Develop Tab view.   In the Library Metadata Tab, there is no mention of the 'as taken' white balance!
This is a Raw file, not jpg.

Can someone please explain this discrepancy between LR and Camera 'as shot' white balance info?  Am i doing something wrong?

Thanks and regards.
  

there is no standard for such things ... WB in raw converters technically is not  some "temperature" & "tint", but either a scalar per raw channel multipliers (for those which do that predemosaick) or (for example in Adobe's case, post demosaicked) matrix multipliers for "RGB" vector (and there might be some raw converters that will operate w/ WB multipliers in non RGB spaces too)... "temperature" and "tint" are presented to you in UI and all rawconverters will calculate (and present) them a little differently (more so with a "tint" part).

so for example if your camera will record WB as a set of scalar multipliers in a raw file then each raw converter will show you "as shot" WB as different "temperature" and "tint" (and some few like RPP will actually just show you per channel multipliers, because that is what it is)... values for the "temperature" part of that "as shot" WB will be generally within the same vicinity (if raw converter can extract that metadata from your raw file)
« Last Edit: August 04, 2013, 11:29:41 am by Vladimirovich »
Logged

eliedinur

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 328
Re: LR 4.4 Incorrect color temperature reporting
« Reply #2 on: August 04, 2013, 06:43:49 pm »

The metadata that LR uses from the file is not the temperature setting either you or the camera's Auto WB set, but rather the actual multipliers for the red and blue channels that the camera determined it would use if it were to make a jpg. But WB is applied after the camera profile and therefore the representation of their neutralizing action as a temperature is dependent on the profile. The camera makes its calculation of the multipliers needed on the basis of Nikon's profile; LR takes those same multipliers and applies them over an Adobe profile, which results in a different translation to a temperature.

This is what Eric Chan wrote on the subject:
"What you're referring to is the As Shot white balance setting, which is the default.  This simply uses the WB values as recorded originally by the camera and written into the raw file.  WB in this case is recorded as a set of gain factors (or equivalently, "camera neutral" values).  The translation between these neutral/gain values and white points (or temperature and tint slider values, which is what you see in the UI) is dependent on the camera profile (color matrix values).  So when you switch between different camera profiles, the translation between the camera-recorded gain values and user-visible temperature/tint values can change, and this is expected behavior." 
Logged
Roll over Ed Weston,
Tell Ansel Adams th

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: LR 4.4 Incorrect color temperature reporting
« Reply #3 on: August 04, 2013, 11:48:33 pm »

This is what Eric Chan wrote on the subject:
"What you're referring to is the As Shot white balance setting, which is the default.  This simply uses the WB values as recorded originally by the camera and written into the raw file.  WB in this case is recorded as a set of gain factors (or equivalently, "camera neutral" values).  The translation between these neutral/gain values and white points (or temperature and tint slider values, which is what you see in the UI) is dependent on the camera profile (color matrix values).  So when you switch between different camera profiles, the translation between the camera-recorded gain values and user-visible temperature/tint values can change, and this is expected behavior." 

Eric talks just about different dcp (Adobe DNG camera) profiles in Adobe products... here is totally different calculations from the recorded info in a raw file to values of temperature/tint sliders presented to you in UI (and the "tint" slider, if present, is graduated completely differently values wise in different raw converters) by different products using potentially different approaches to when and how and where to apply WB...

PS: matrices used for WB are an integral part of a dcp profile so it is not a proper way to say that "WB is applied after the camera profile" in Adobe's case.
Logged

André Denis

  • Newbie
  • *
  • Offline Offline
  • Posts: 17
Re: LR 4.4 Incorrect color temperature reporting
« Reply #4 on: August 05, 2013, 09:43:32 am »

Thanks guys,

Knowing how the WB, expressed as a standard light temperature in K, is implemented internally by different vendors is certainly an interesting topic by itself.   But my question and concern is really about traceability and acquiring experience.   
I take thousands and thousands of pictures, and even thought  I shoot in raw, and i know i can later do some WB adjustment, i try to make the best settings when I take the pictures.  Old reflexes are hard to discard. WB expressed K is my reference and slowly, over the years, I gained an ability to judge the color of the light, in K, and set the camera accordingly.   Now, I understand (and see) that the final result can be very different depending on the camera profile, but also on the vendor of the raw conversion application (LR, NX, C1, etc.) i use.  Nevertheless, the world have to make sense and be stable so that we can build knowledge and when I set the camera to 7000K, I want to be able to know one year after, or one week after, what was the K I had set in the camera (the light might be the same, or slightly different).  It seems that contrary to other vendors,  Adobe Light Room designer has chosen not to display my original WB intent (in K), but the internal numbers they use to achieve that WB.   Suddenly, 7650K -NIKON  is the same as 7400K - in LightRoom.  Its a programmer's playground, not a photographer. I take pictures and don't run around with conversion algorithm (even though once in a while its fun to talk about it).

So is there a way... a hidden button in a nth level, a secret password, where LR can remind me what were those 4 digits i used for K when I took the shot? Everybody else does it.  It would be very useful for me.

Thanks.
Logged

Tim Lookingbill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2436
Re: LR 4.4 Incorrect color temperature reporting
« Reply #5 on: August 05, 2013, 12:51:41 pm »

Quote
So is there a way... a hidden button in a nth level, a secret password, where LR can remind me what were those 4 digits i used for K when I took the shot? Everybody else does it.  It would be very useful for me.

Line up all your previews of the same image opened in the "Everybody else does it" crowd in the same RGB output space and compare to Adobe's and use the same K numbers between all of them and see if you get a match in their appearance.

I'm betting you don't from what I've experienced doing the same with a few other image editing apps that can preview and render Raw images.

You should be more concerned by how the WB looks with regard to its nuance and ambience it imbues to the overall look of the image, not the numbers. Your camera and Raw converter aren't designed to be precision scientific imaging devices. They're designed for making pleasing looking images which is a subjective based concept and is the reason they all include adjustment sliders and other editing tools.
Logged

eliedinur

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 328
Re: LR 4.4 Incorrect color temperature reporting
« Reply #6 on: August 05, 2013, 01:30:45 pm »

Quote
So is there a way... a hidden button in a nth level, a secret password, where LR can remind me what were those 4 digits i used for K when I took the shot?
Not LR, but a specialized metadata reader like ExifTool. It is the fourth field in the Nikon Maker Notes. There is a ton of information in the Maker Notes, but most of it is irrelevant in the context of what LR does and there is no reason display it.  There is also far more information in the EXIF than LR displays, but to show it all would be absurd and LR, like every other photo editor, cherry-picks only the fields that the designers think most users want to see. The good craftsman uses the tool that is appropriate to the task and there is readily available software that will show you all the metadata.
Logged
Roll over Ed Weston,
Tell Ansel Adams th

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: LR 4.4 Incorrect color temperature reporting
« Reply #7 on: August 05, 2013, 02:47:29 pm »

So is there a way... a hidden button in a nth level, a secret password, where LR can remind me what were those 4 digits i used for K when I took the shot? Everybody else does it.  It would be very useful for me.

Thanks.

only if Nikon writes that into metadata in "Kelvins" in addition to WB multipliers and LR has some plugin to display that metadata (exif tags).
Logged

André Denis

  • Newbie
  • *
  • Offline Offline
  • Posts: 17
Re: LR 4.4 Incorrect color temperature reporting
« Reply #8 on: August 05, 2013, 04:05:00 pm »

I understand the answer is no... :-[ Too bad.

We all agree that in the end its what the picture looks like that counts.  But it's also a process and a learning process, trial and error.   Discarding information is, for me, not a good thing.  I'll probably keep using LR, because the tools are there to tweak the image the way I want, but not the tools to help me choose the right K when I shoot, which i'd like to do ... also.  I know I know, most of you will say 'Who cares, its in raw!'.  True, but I care.  :)

Yes aligning all versions of the different vendor will yield a different color rendition.  This is why i use profiles, calibrate my screen, and not switch constantly between vendors.  I want to learn one set of tools.  LR is new to me. I just won't be able to feedback it's K information back to the camera. 

When I look at my pictures, after a run, i don't just look at the images but also at how i set the camera. Where did i do good and where did i do wrong. If the color is off, of course i can correct in LR, but i also want to do better next time,in the field.  And for that, having the in camera K is useful.   I'll also keep PhotoMechanics.

Thanks all for your reply. 

Regards
Logged
Pages: [1]   Go Up