P: ORF white balance inconsistent

Hi there,

Is there any update on the issue? Seems like LR is still unable to render consistently images from a sequence. It's very problematic for timelapses when tint changes every few frames.

Thanks,
Sergei

Thank you Melissa!

Here's another recent post with the same 'Black Level 2' White Balance shift in two virtually identical ORF raw files. https://forums.adobe.com/message/9292439#9292439

As Steve Sprengel mentioned the issue appears to be that LR does not properly use the .ORF file's 'Black Level 2' setting . See this post for more details:https://forums.adobe.com/message/9100913#9100913

Hi guys,

I hope I didn't step on anybodies toes and if so, I am sorry.

I just would really like to understand this (a scientist's weakness ;)) and find, or help to find, a solution either in my work flow or for Lightroom or for Olympus.

Right now I'm still in a limbo, not really knowing how to proceed. I love the Lightroom work flow and would really like to keep using it.

Cheers
Stephan

Hi guys and thanks for your replies.


@273939: They are definitely different. If you import the files to LR and push up the exposure by 1-2 stops, you'll see a clear difference.

@796284: I'm not an expert on image files, so I'm not fully sure how to interpret your posts, but I think you are saying that the pictures are different and LR is treating them consistently. This is fair enough, but does not solve my problem, since:

I'm not arguing that LR is doing something intrinsically wrong. I'd rather say that it is doing something different then the Olympus software and that the latter gives "correct", in the sense of consistent, results.
I created four more files and uploaded them to https://pitufos.de/ORF/. The jpegs with the original file names are produced via:
ORF-file -> Import to LR -> push exposure 2 stops -> export as JPEG
The second one appears much more yellowish. Looking at the histogram, it seems like the blue curve is shifted to lower brightness.

The second pair, named a and b, respectively are produced via:
ORF-file -> Open in Olympus Viewer 3 -> export as TIFF -> import TIFF in LR -> push exposure 2 stops -> export as JPEG
This pair looks identical to me. But I consider this method a workaround at the best.

I also tried dcraw to process the files and couldn't find any differences in colour either.

In summary: Even though it might be that the ORF files are different, there must be something (maybe the black level value) that allows the Olympus viewer and dcraw to create consistent results. Therefore, even if LR is not technically doing something wrong, I would argue that the result is incorrect. If that is due to a problem in ORF, then I guess LR needs to work around this and do the same "magic" that the Olympus Viewer does to give consistent results.

Side note: I don't have a problem if different raw converters give slightly different colours in the end. But they should give consistent results in a series of pictures of the same scene. And this is not entirely random when using LR. P...73.ORF looks consistent to P...71.ORF and so do the next 5 pictures. And then come another three, consistent with P...72.ORF and then back to P...71.ORF and so on and so forth... . Quite frustrating.

Regards
Stephan

The histograms are from the actual raw data pixel values in the raw file w/o any adjustment or WB being applied.  There are two green histograms because there are twice as many green pixels as red or green.  Since the camera is reporting a different black point and the histograms are slightly different especially at the dark end, the images are different. 

Here is a difference of the two histograms done using the Color layer blend mode of PS:


Between EV -6 and EV -5, the histogram is showing the number of pixels at approximately each 1/8th EV brightness level.  The difference from one histogram peak to the next in the EV -5 to EV -6 range is about the same as the difference between 71 and 72 so the brightness difference between the two images is about 1/8th stop in the darker sky area.  A specific example is that P...72.ORF  has the same number of EV -5 6/8 pixels as P...71.ORF has at EV -5 5/8 and the trend continues from EV -7 to EV -4 or maybe more:

They certainly look identical for me. Are you sure that you haven't somehow changed the color temperature between the two? They come in for me as 5,200° +12, and they look the same in both Photoshop and LR.

The first image:  P...71.ORF is slightly brighter in the darkest areas so the color balance in the darkest areas is slightly different given the different weights the R, G & B channels contribute to the final luminance value.  Basically the first one is bluer and the second one is greener, but not because LR is messing up but because the pixels values are very slightly different for who knows what reason.  The camera may be suppressing color differences more in the darkest areas.  What I'm not sure is if the Black Level value stored in the EXIF of each photo is something LR uses or not. 

Here is a side-by-side of a portion of the EXIF information as reported by the website:  http://regex.info/exif.cgi


And here are the histograms from each image in RawDigger where you can see the portion between EV -7 to EV -3 or so is slightly higher in the 71 image vs the 72 image: