conversion from SER to FITS loses Bayer-pattern information
Siril version: commit c91f81dd
Note: please report against latest git master branch or latest stable version
Operating System: Ubuntu 20.04
Description of the bug
Conversion from SER to FITS loses the Bayer-pattern information.
Reproduction
Reproduction steps:
-
Siril has Bayer-pattern set in the preferences as RGGB, same as the SER-file. ("Top-down" and "use from header" is enabled.)
-
Use a SER-file with RGGB Bayer-pattern and convert it to FITS in the Conversion-Tab. Do not enable Debayering because it's not possible yet (BTW: Why is it not possible?!)
-
Load the converted FITS and Debayer it in the Conversion-Tab. Log shows:
No Bayer pattern found in the header file.
Filter Pattern: RGGB
Saving FITS: file zz_00001.fits, 3 layer(s) ...
- Load the debayered FITS into Siril, zoom in and observe the maze-pattern structure due to wrong Debayering.
Actual result:
Debayering is apparently done with a wrong mask (it seems the image is saved or processed in the wrong order as the mask GBRG produces correct looking results).
Maybe it is enough to store the correct Bayer-mask into the FITS, apparently this is not done.
Also it is mysterious why the image is possibly processed in the wrong order (in case if the correct Bayer-mask is used).
Additional information
The information should be enough to find the problem, however I can provide test-files to reproduce the bug, if wanted.
Thanks.