PureRaw 3 messes up Greens with Olympus OM-1

Here are the basic PhotoLab 6 versions with the two possible color renderings available in the generic rendering:

  • DxO Camera profile (OM-1)
  • Neutral color

Note: files in Adobe RGB color profile originally, converted to sRGB. It is possible that there is a small variation compared to the originals seen on Adobe RGB screen.


Of course, no special settings have been made in PhotoLab. These are identical to PureRAW, except of course for the neutral color profile which is not possible in PR.

What is shown here is therefore a small part of the possibilities of PhotoLab.

Regarding the white balance, it should be understood that there is no absolute scale of values with regard to digital photography. It may sound strange, but it is the reality.

Indeed, and contrary to what you think, the white balance is not recorded directly in the exifs. The only WB indication present is of the “Auto, Custom” type. The sensor does not record according to a given white balance, it records in its own color space. Which is able to restore during demosaicing, to a certain extent, all of the possible combinations of white balance. In the camera, the white balance that is defined (whether manual or auto) is only used for one thing: to display correctly (but not necessarily exactly) on the screen/viewfinder, and to take out a jpeg camera if this option has been chosen.
There is absolutely no reason to assert that the true value of WB is that of the manufacturer: it is his and everything is fine as long as we stay in his ecosystem: camera + internal demosaicing + demosaicing with his software provided.

Like each camera manufacturer, and each demosaicing software publisher uses its own white balance scale. Do not look for identical values between the camera, and what the dematrixer x or y displays! The third-party software knows how to retrieve the values from the raw file data, and transpose them into its own system.
For example, for the color temperature of your image which I suppose is set in the camera to 6300 K, (if I believe the title of the file, but we do not know the value of the tint), we find in correspondence the values temperature and tint of the raw file:

  • 6800 , -12 in ACR (PS, LR)
  • 6664 , -11 in PhotoLab
    If I start from the linear dng produced by PureRAW (or PL6), the values are ACR (PS, LR):
  • 7000 , -12
  • 6990 , -12

But these differences are almost irrelevant, since in practice the result compared is very close if not identical. For example as shown above, the colorimetry between the raw and the dng opened in ACR is visually identical while the values are different. It doesn’t matter if x or y scale is used, what matters is that what is taken from the raw file during demosaicing is similar in terms of color.

Thank you very much for the images. It’s interesting to me the one produced using DXO’s OM1 profile is even further away from the camera produced jpeg than the DXO Neutral profile. I’m not sure what they were trying to achieve with the OM1 profile, but evidently it was something other than matching the manufacturer’s colors.

In your eagerness to find ways to correct me, you keep doing it with things I’m not mistaken or confused about.
Indeed, and contrary to what you think, it is you who are very much mistaken about what is in file. Please feel free to use exiftool to dump the info from the file. You will find that “6300” is contained in the “WhiteBalanceTemperature” field. Perhaps Photolab doesn’t know to read that field.

Then there’s also the WB_RBLevels field, which PL6 is probably interpreting, like ACR does, to convert to its own idiosyncratic white balance parameters.

So no, the only WB indication present is not unknown or custom or manual, and certainly not Auto. You won’t find tint recorded by cameras either because it’s not proper parameter like Duv. It’s a made up construct created by software vendors and isn’t defined by CIE.

But thank you again for the images and for clearing up some questions I had about PR3 and Photolab. That was very helpful. It’s disappointing that PL6’s color processing for my camera does not better meet my needs. Hopefully in the future it will improve.

Cheers, and thanks again…

Response a little late, but I have been held back by other occupations for the past two days.

I would like to add a few points on white balance:

  • In photography, the White Balance is always defined by two values: The color temperature in °K (yellow/orange/amber - blue axis) and the tint in values derived from CC (magenta - green axis).

  • When I write that the WB is not registered in the EXIF, it is indeed the case! The correspondence with a WB of 6300 that you find is NOT in the EXIF (which is perfectly “regulated” metadata), but in the “open” MAKER section, where the manufacturer records the data he wants, including those which will be used to create jpegs identical to the camera jpegs in his own demosaicing software.
    Most of the data in this section is read and can be used by third-party software.
    Others have no interest outside of the brand’s proprietary demosaicing software: only it knows what, for example, data concerning “picture styles” or other such customizations correspond to.

  • In the case of the white balance recorded in MAKER, it is a data which is only applicable in the precise scale of values of OM, and which most probably combines the values of color temperature and tint.
    Third-party software (PhotoLab, ACR, etc.) does not use it: this value is not relevant for them.
    As you have noted, in this MAKE section are given the actual values (in a perfectly defined reference frame) of the RGB levels for the various typical and custom WB settings of the OM-1. In the case of the custom WB of your raw, this value is: RBGG = 618 394 256 256.
    For third-party software, with these RGB ratios and color matrix info, there is no problem to reconstruct the WB (= color temperature + tint) in their own scale!

  • It does not matter that the TC and tint do not correspond exactly between software, the basic data is the same, and the rendering is therefore almost identical.
    And by the way, of course OM sets WB and Tint values, like absolutely all brands, and of course you can adjust the tint in the OM-1 with the White Balance Compensation setting!

  • And finally it is inaccurate to say that Tint is a creation of software publishers, for the very simple reason that tint existed long before PhotoLab or Adobe: all thermocolorimeters before digital (and therefore before demosaicing software! ) gave the values of Color Temperature AND tint (value of CC = Color Compensation )

Hello all, I just found this post.
I am seeing similar effects with my new (used) Lumix G9.
I have used PR3 with my Lumix G80, FZ10002 and FZ82 with no problems so I was very surprised to see this with my G9.
Has anyone else seen this?
And is there a solution?

PS: When I import the raw files into SilkyPix I do not see the colour shift - the generated jpgs are similar to those generated in camera.