Colorbar labels overlap in Brain plot without time viewer

When plotting an STC without the time viewer, the colorbar is arranged horizontally, causing the individual labels to overlap depending on the window size:

# %%
from pathlib import Path
import mne

mne.viz.set_3d_backend('pyvistaqt')
sample_dir = Path(mne.datasets.sample.data_path())
stc_fname = sample_dir / 'MEG' / 'sample' / 'sample_audvis-meg'  # omit suffix
fs_subject = 'sample'
fs_subjects_dir = sample_dir / 'subjects'

stc = mne.read_source_estimate(fname=stc_fname, subject=fs_subject)

stc.plot(
    time_viewer=False,
    size=(300, 300),
    subjects_dir=fs_subjects_dir
)

I’d suggest to only label the anchors and midpoint.

cc @GuillaumeFavelier

❯ mne sys_info
Platform:       macOS-12.1-x86_64-i386-64bit
Python:         3.9.7 | packaged by conda-forge | (default, Sep 29 2021, 20:33:18)  [Clang 11.1.0 ]
Executable:     /Users/hoechenberger/miniforge3/envs/mne/bin/python3.9
CPU:            i386: 8 cores
Memory:         16.0 GB

mne:            1.0.dev0
numpy:          1.21.4 {blas=NO_ATLAS_INFO, lapack=lapack}
scipy:          1.7.3
matplotlib:     3.5.0 {backend=MacOSX}

sklearn:        1.0.1
numba:          0.53.1
nibabel:        3.2.1
nilearn:        0.8.1
dipy:           1.4.1
cupy:           Not found
pandas:         1.3.4
pyvista:        0.32.1 {OpenGL 4.1 INTEL-18.3.5 via Intel(R) Iris(TM) Plus Graphics OpenGL Engine}
pyvistaqt:      0.5.0
ipyvtklink:     0.2.1
vtk:            9.0.3
PyQt5:          5.12.3
ipympl:         0.8.2
mne_qt_browser: 0.1.7
pooch:          v1.5.2

[larsoner]

Personally I’d also like it to be easy to set the colorbar tick values/locations manually…

Fun fact, this wasn’t even possible in even the dev version of VTK until 9 months ago, mostly (I think?) because ParaView had their own subclass that did something similar. Fortunately it does look like VTK 9.1 included the change that allows doing it:

>>> brain = stc.plot(...)
>>> from vtk.util import numpy_support 
>>> numpy_support.numpy
>>> import numpy as np
>>> brain.plotter.scalar_bar.SetCustomLabels(numpy_support.numpy_to_vtk(np.array([4, 5, 17], float)))
>>> brain.plotter.scalar_bar.SetUseCustomLabels(True)

To me the best place to make all this accessible would be to subclass vtkScalarBarActor and implement at the PyVista backend as I mentioned above. The vtkScalarBarActor is currently accessible via brain.plotter.scalar_bar and brain._scalar_bar, so I say we change it to brain.scalar_bar (public name) and then make brain.scalar_bar have nice methods at the PyVista level. It’s much more in PyVista’s scope to provide convenience wrappers for VTK “stuff” than MNE-Python, so I’d rather work to solve the problem / implement a solution at their end with the help of their expertise!

[drammock]

Maybe we can discuss some ideas on Friday? 😃

Sure

Do you see this problem in your daily work?

Not daily, but it does come up, and it does annoy me. Personally I’d also like it to be easy to set the colorbar tick values/locations manually… e.g. if I’m plotting negative log10 of a p-value on the brain, then certain specific values are interesting / important to signpost