[stevebrot]

What is remarkable is that on Windows 10, when I open such file in the standard viewer, it is also yellow, but not in RT or camera raw. Can you think of what reason this could have?

It depends on the codec. Most RAW image codecs simply extract the thumbnail JPEG packaged with the RAW for use in the file browser. This is done with no RAW processing by Windows and for Pentax is the similar to extraction of a larger preview for is shown on the rear LCD. I will have to go back to my notes from a few years back, but if my memory is correct, the DNG for pixel shift packages the preview thumbnail JPEG for image #1 of the four component DNGs with no pixel shift processing having been done by the camera at all. A similar approach is used for the image viewer, only using one of the other JPEG preview images (there are several including one at full resolution).


Steve

[stevebrot]

should have the same WB properties

Indeed, though if the capture light is different, the processed image (what we see on the preview) would reflect that difference. That is why an actual extraction of the JPEG preview images from each of the four component DNG would be useful.

Be sure that I am not convinced that what you are seeing has an easy answer. I suspect something subtle is happening that is hard to figure out without having two people sitting at the same computer brainstorming.


Steve

[Svenneke]

Hi Steve,

As soon as I find some time this week, I'll do the test and will try to extract each of the four images.

With 'problems' like this, I would not be surprised at all Steve if the answer should be something rather simple.
It sometimes only require one single parameter between the whole series of calculations that is wrong or has a range that falls beyond what was required, and the end result could become unexplicable. I often make rather complex Excel files and use Access daily for work and sometimes typing one simple error in a formula can make the whole spreadsheet useless.

One day I had a problem in Access and I could not search for new entries anymore in the database. It took me 2 days to find out what happened. The IT-specialists who are supervising our computer systems couldn't find the answer neither.
I persisted and found out myself that by entering a date in a field and pressing "enter", instead of going to another reference field, the database became corrupted because that "enter" changed another value as well in a reference table that it was not supposed to do. It literally changed a value 'behind the scenes'. In a reference table the 1 in that field caused the database to look at what was given as input by the user and searched the corresponding data.
After the 1 became a 0 the database no longer looked at the data that was asked by the user, but only to the last data that was used before, and continued to do that forever. This caused that I could not search for new entries anymore and was a real PITA.
It took me almost two days to find out that a one in the background was replaced accidentally by a zero, only because of a fault in the way the database was constructed. I changed that zero back to a one, saved the changes and immediately everthing worked again as before.

In the case of the K1 an the weird yellow color under those specific circumstance, I would not be surprised if something similar simple happens.
But of course it also could be something subtle and at the same time very complicated like you suspect.
It would be nice to sit together on the same computer, but unfortunately the distance between us is a little problem ;-).

Have a good day Steve, I'll keep you up to speed if I have found something.
Greetings,
Sven