Fix: nearly invisible traces with spatial_colors=True

[Mojackhak]

Reference issue (if any)

Fixes #6686
Fixes #13285

What does this implement/fix?

This patch modifies the _rgb function in mne/viz/evoked.py to reduce the intensity of overly light RGB colors that can occur when spatial_colors=True is used in PSD plots. Specifically, it checks whether the sum of RGB components is above a threshold (2.5) and reduces the brightness to improve trace visibility.

This issue was previously discussed in Discourse thread #6572, and this fix follows the workaround suggested by @bboyth, which has been confirmed to work in MNE 1.9.0 by multiple users.

Additional information

None

[welcome]

Hello! 👋 Thanks for opening your first pull request here! ❤️ We will try to get back to you soon. 🚴

[larsoner]

Can you add a doc/changes/devel/13286.bugfix.rst with your change using the :newcontrib: directive (see other examples in the doc/changes/devel folder)?

And then can you make some trivial change to https://github.com/mne-tools/mne-python/blob/main/tutorials/evoked/20_visualize_evoked.py ? Any trivial change should work (e.g., changing whitespace) as it will make CircleCI rerender the example so we can see if it changes

[Mojackhak]

thanks for your patience! I’ve pushed the updates.

[larsoner]

One more thing from above ☝️

And then can you make some trivial change to https://github.com/mne-tools/mne-python/blob/main/tutorials/evoked/20_visualize_evoked.py ? Any trivial change should work (e.g., changing whitespace) as it will make CircleCI rerender the example so we can see if it changes

[Mojackhak]

sorry for that. I've pushed another update

[larsoner]

I looked at the example and the images didn't change, which I guess is okay since none of those were really invisible. Adapting the example from one of the issues above:

import mne
from mne.datasets import sample
data_path = sample.data_path()
raw_fname = data_path / 'MEG/sample/sample_audvis_raw.fif'
raw = mne.io.read_raw_fif(raw_fname, preload=True)
raw.drop_channels(raw.info["ch_names"][:315])
raw.drop_channels(raw.info["ch_names"][4:])
raw.info["bads"] = ['EEG 003', 'EEG 004']
raw.plot_psd(spatial_colors=True, picks=range(4), exclude=())

I do see one channel that was hard to see get a bit darker

main	PR

In the future we might consider using some other smoother transformation of all lightness values (like a sigmoid or even just a linear rescaling) rather than a threshold-based one since this one could lead to color collisions but this already seems like a usability improvement.

So I think this is good to go, marking for merge-when-green, thanks in advance @Mojackhak !

[welcome]

🎉 Congrats on merging your first pull request! 🥳 Looking forward to seeing more from you in the future! 💪

[Mojackhak]

Thanks for double-checking and merging! 🙌