adds a slider for the head surface opacity (it effectively replaces the previous head_transparency and its automatic opacity management)

I don't think we should get rid of some of the opacity management. I think that whenever the fiducials are unlocked, we should go to a fully-opaque state. (It's okay to let the user change it from this, but I doubt they will.) Once they lock the fiducials, it should go back to its previous opacity (or to the default opacity if the UI launched in edit-fiducials mode).

fixes the range of the x/y/z combo boxes

Make sure you fix the step, too -- I'd go for 1 mm, 1 degree, and 1 percent for the translation, rotation, and scaling, respectively.

Let me know when I should test!

How is it now @larsoner ?

correcty updates the plot when the trans parameter is set

The plot, yes:

mne coreg -s sample -d subjects -f MEG/sample/sample_audvis_raw.fif --trans MEG/sample/sample_audvis_raw-trans.fif

The rX/rY/rZ, no -- they should not all be 1. Did you fix the range of these so that they are always >= 1? I think these actually in practice go from something like -180 to 180, but please do check the range. It could even be -360 to 360 degrees, not sure...

The translations similarly seem fixed to >= 1, when they should be allowed to range at least -100 to 100 (mm), maybe even -500 to 500 just in case someone's MRI has a very weird affine.

The range of 1 to 1000 percent for scaling seems reasonable, though.

In fixing a test I broke flake, so please fix this while fixing the above stuff:

Error: ./mne/gui/tests/test_coreg_gui.py:178:80: E501 line too long (81 > 79 characters)

• The slider behaves in an odd fashion: there appears to be some kind of a huge step in the beginning, before transparency can be adjusted smoothly (see video below)
• I thought we had discussed not to add a slider (or any transparency adjustments) if we figure that we can live with a fixed opacity level of 95% – am I mistaken here?

am I mistaken here?

I think you voted against a slider but @drammock and I voted for a slider

The slider behaves in an odd fashion

Hm... I believe this is how vtk handles transparency. I have the same on my linux. I would like to have a smoother transition too

@GuillaumeFavelier should I test again? Did you test locally a bit to make sure these things worked properly?

Make sure you fix the step, too -- I'd go for 1 mm, 1 degree, and 1 percent for the translation, rotation, and scaling, respectively.

I did not check this yet.

I still have to deprecate head_transparency in favor of head_opacity (after #10222)

I also found a bug with mark_inside when using scale mode "uniform", all the hsp become black as if they are detected inside the head surface:

@GuillaumeFavelier @larsoner i will also test tonight (currently out for a walk) and can report back!

After extracting some code from it, now this PR:

• renames the MRI groupbox from "Transform" to "HEAD <> MRI Transform"
• brings an immediate solution to ICP "spamming" by disabling all the fit buttons
• fixes the range of the x/y/z spin boxes
• correcty updates the plot when the trans parameter is set
• fix the x/y/z scaling spin boxes increment
• improve the UX in scale mode "uniform"

I fixed the conflicts, can you test it @hoechenberger ?

62ae738 also fixes:

Aligning using ICP
Start     median distance:  17.40 mm
  ICP  1  median distance:  12.46 mm
Traceback (most recent call last):
  File "<decorator-gen-563>", line 24, in _redraw
  File "/Users/hoechenberger/Development/mne-python/mne/gui/_coreg.py", line 692, in _redraw
    for key in self._redraws_pending:
RuntimeError: Set changed size during iteration

This is ready for review @larsoner, @hoechenberger

@hoechenberger as per your offline review, I added 4ec8b39

Closing in favor of #10283 and #10276