Hi there,
I found that Darktable won’t read the white balance from my Panasonic GH5 raw images.
You can see from the image that the white balance field is empty.
In Darkroom the Color Calibration module defaults to 4210K even when set to “as shot in camera”.
Since it’s an already “old” camera and quite well known and worked with, I thought this might be an unintended issue.
The RAWs are the original RW2 files straight out of the camera and have not been converted.
What do you see without CC module and as shot in WB…does it look right. Does it match your file if you look in the meta data?
Don’t use the CC temperature as a reference. It’s not going to translate…it’s well explained in the manual
Yeah I think it matches the RAW’s embedded preview in some other pics.
I had a scene lit with colored lights and with Filmic RGB the colors were a bit off. By adjusting CC I got it closer to what it was it was supposed to be but after watching a Sigmoid tutorial I think my issue was Filmic RGB and how it treated colored highlights. So I’ll stick with Sigmoid and learn from there.
Can you at least treat these CC values as white balance by doing subtractions/additions?
Let’s say I shot at 5600, CC says 3622, if adjust CC to 2622 does this translate to 4600?
Basically, ignore them…use the illuminant color in the patch for your reference…CCT are being reported but if you look at the perpendicular lines on the Plankian (black body) curve each point on that line has the same CCT despite being quite different in color….
This is where the invalid tag comes in …if the CCT is too far away it cannot be modelled by daylight which is a similar curve but just shifted up a bit or the black body curve. So then you see DT will shift to the hue chroma interface and completely drop temperature.
CC module is not really the same as white balance which is considered a partial chromatic adaptation….
There is some good information here
https://www.waveformlighting.com/tech
Esp
https://www.waveformlighting.com/tech/calculate-color-temperature-cct-from-cie-1931-xy-coordinates
And here
https://www.waveformlighting.com/tech/calculate-cie-1931-xy-coordinates-from-cct/
DT is finding the xy chromaticity co-ordinates of the illuminant and then doing a CAT to adapt that so that the colors look the same under the viewing conditions illuminant… so I believe taking it to D50 white of the pipeline. This is a matrix calculation so that’s why I believe it lands in the channel mixer module. You can see this in action sort of when you select the option in CC for same as the pipeline and you will see a white patch and CCT of like 5002 or something like that…
The process and principles not explicitly for how DT does it but in general are well explained here
https://yuhaozhu.com/blog/chromatic-adaptation.html
and
Here I was saying if you look into the exif data the file should state the white balance used and generally also you get the wb coefficients used for a standard WB… I was just asking if they seemed to match… I use xnviewer and it uses exiftool as a plugin so its easy to review the meta data … there are lots of other ways command line and other to view the complete or most of the readable data
This video by the developer has a lot of good information on the subject:
I just loaded the test image from DPReview…
Exif
Turning off CC and setting WB to as shot and doing the math…the scale values are read and correct…
As for the CC reporting and CCT there are explanations provided above…In the case of this image this is what DT reports in WB
No. See here:
