Hi all I've just recently come back to using my DSLR and noticed an oddity with the FITS images produced by KStars/EKOS.
Compared to the CR2 file the FITS image produced seems to have lost some data in the conversion somewhere between the camera and landing on my disk through EKOS.
When captured as "Native" it come out as expected. However in FITS its has all turned green!
Yes I have turned of auto stretch and checked all my settings, RAW format and image dimensions. See attached images when pull into PixInsight below.
The FITS has been debayered in PixInsight with VNG and CFA GBRG and represents exactly what I see in the EKOS/KStars Fits Viewer.
This is a link to the raw files for comparison on OneDrive: 1drv.ms/u/s!AvWEkn9Anb_NgU5r_vS7LLTTDgeS?e=xFcXtZ
I figure some update to gphoto2 or some regression on how kstars/indi/ekos interprets the raw stream from the camera?
Any ideas would be appreciated. Thanks.
Last edit: 2 years 11 months ago by Maria. Reason: Added extra info.
Do the KStars settings for FITS have an option for auto-debayer? If yes, is it enabled? If yes you may want to turn that off. Not sure if the setting affects images written to disk or only the FITS viewer though.
I've tried disabling the auto debayer, and the result is just a non-debayered image in the Fits Viewer, and it doesnt change the resulting file.
I've edited my original post and added a link to the raw files in onedrive if you want to take a look.
Thanks for the samples. There indeed is a difference between the two images. The FITS version has 16-bit channels while the CR2 version is pure raw data with 8 bits per channel. The histogram of the FITS data has been severely clipped as can be seen in this screenshot of APP:
The CR2 data has a much better histogram. It looks clipped on the right but that's because it is overexposed at the center:
So I suppose the conclusion is: shoot your images in the raw data format of your camera. In other words, shoot in CR2 and not FITS to avoid data loss. All modern image processing applications handle CR2 data very well so there really is no need to shoot in FITS anyway.
Yes indeed I'll be falling back to cr2, although the intent of raising this here was to hopefully highlight a potential bug. It might be something that could be fixed and could be impacting others.
Is this something we can raise with Jasem?