A month ago I reported missing keyword for Filter in the FITS header. I shot more flats today and the same thing occurred - some files (a majority) were missing the Filter keyword in the header. (Causing WBPP to sort them as "no filter".

There was no response to my last post on this topic. Am I the only one? My flats are saved to my indoor laptop (client) rather than the Pi on the telescope. Camera is a QHY268M with the QHY FW3 filter wheel. Filters change as expected, filter name gets written to filename as expected, but only intermittently to the FITS header.

Anybody have any ideas on this? In this case all the exposures were short, in the neighbourhood of 0.05 to 0.12 seconds. (new flat source much brighter than expected - still experimenting)

This was with an updated Kstars - 3.5.8. Stable, Build: 2022-03-24T22:52:45Z

Same experience here. Seems to be limited to Flats. But strange thing is that it is not constant for one filter. For each filter, some files have the Filtername written in the FITS header, others don't.
I am still on 3.5.7, but did not have this problem in the past.
Using KStars on Mac, exposures are 2-10 secs.
Like you say, very annoying, because you now have to calibrate the flats manually before you can use the automated WBPP script.

Willem, I noticed this issue first with Flats and 3.5.7. I updated to 3.5.8 three days ago, just before a rare clear night two nights ago. I noticed the issue when PixInsight's WBPP script reported a number of Flat exposures fell into a "No Filter" category. This time a few light frames were effected too. Filenames properly show filter used but FITS header is missing the Filter keyword. Last time I think only Flats were involved.

I'm hoping for a resolution as individually adding the proper keyword and value to each file is time consuming. I haven't the skills/experience to write a script to automate the process.

David, yes, what you describe is exactly the same behaviour as what I experience as well. It looks like it is an error introduced in 3.5.7 then. And I use ZWO filterwheel, run KStars on a MacMini (Mojave) and store files locally. Since our setup is quite different, it seems unlikely that this is driver or OS related. Hopefully this will be picked up and corrected in future updates. I will try to see if I can post this as a bug.
In the meantime I have been doing some searching on the PI forum and found a nice workaround. Here in this thread they show other examples of acquisition software not writing FILTER keywords to the FITS header. Towards the end of the thread, someone has created a script to do this on whole batches of files at once. I have just installed that script and it works very well. My WBPP is currently crunching a whole bunch of files, many of which have been edited using this script. Not ideal, but a workable workaround.
Oh, and the script works so familiar, that it might open up possibilities to use FITS-header info even more extensively. For your convenience, attached is the script file.Let us know if this works for you too.

Willem Jan.

I've posted this as a bug on Github: github.com/indilib/indi/issues/1649