Mihai Cara

I do not see any obvious issues with `tweakreg` itself. One thing to check is to look at the log file and see which of the input images was chosen...

Try sorting input file list to ensure the same order. See if this helps improve consistency.

CC: @jhunkeler

I would say that I just want to improve one aspect of existing notebooks to address a specific issue. If there is desire to re-design the notebooks, that could be...

[`gwcs.WCS.to_fits_tab()`](https://gwcs.readthedocs.io/en/latest/api/gwcs.wcs.WCS.html?highlight=to_fits_tb#gwcs.wcs.WCS.to_fits_tab) could be used to simplify generation of `-TAB` FITS WCS.

> Of course it also can't really hurt to have it... It can, if provided inverse SIP approximation is not very accurate due to how DS9 has been set-up to...

On the other hand, current level of accuracy (0.005 pixels) for the inverse is high enough for visualization applications (DS9) and `astropy.wcs.WCS.all_worls2pix()` ignores inverse SIP. So does AST (by default)....

Again, I recommend not to write out inverse SIP but I understand if this is desired and that is the only reason for this PR.

@oxno2's image (and possibly @cloud182 issue) suggests to me that the new version has a bug resulting in a diverging algorithm somewhere. I disagree with the main points in https://github.com/astropy/photutils/issues/951#issuecomment-537707417...

@SaraOgaz functions to work with bitmasks have been in ``stsci.tools`` for quite some time: https://github.com/spacetelescope/stsci.tools/blob/master/lib/stsci/tools/bitmask.py ```python >>> from stsci.tools import bitmask >>> import numpy as np >>> dqbits = np.asarray([[0,0,1,2,0,8,12,0],...