BUG: TFR.plot misc issues

I am experiencing several issues with the .plot method of TFR objects. I suspect that there are several things off, but for now I found this obvious bug:

These are valid inputs for the title parameter:

https://github.com/mne-tools/mne-python/blob/638847ea5539be32919b6ac4dd85ef8c4c3b79aa/mne/time_frequency/tfr.py#L1136-L1139

Yet in the code, we have this logic:

https://github.com/mne-tools/mne-python/blob/638847ea5539be32919b6ac4dd85ef8c4c3b79aa/mne/time_frequency/tfr.py#L1269-L1278

Hence, when I pass title='auto', my figure will have the title auto

[eort]

oh boy, this is really quite a challenge to unpack 😃

a couple of design question:

The auto value for title, number of picks, and combine are co-dependent

title='auto' is supposed to concatenate channels names for up to 6 picks. This would only make sense when there are multiple picks and those picks are combined into a single figure (with the combine keyword). Otherwise, I don’t see why I would want to have all picks in the title of a figure that only shows one pick. In case multiple picks are combined, I am not sure whether listing the channels is the best representation. At least in viz/utils.py it seems to be to produce a title like RMS/AVG of X electrodes instead of just a list of channel names. Perhaps a combination of the two would be better? (e.g. AVG of Ch1, Ch2, ..., for fewer than 6 picks).

In any case, what mne promises auto will do, is only reasonable for a specific combination of parameters. What would be the best approach to deal with that? Explain it in the doc? Raising exceptions? Silently adjusting the output whenever it does not make sense (i.e. multiple picks, uncombined)?

return separate figures or subplots?

Right now, the function promises to return multiple figures (even though it actually only returns a single figure). Is the idea to have a separate figure for each pick, or would it make sense to collect the tr plots in subplots of a figure and return that one? Or allow for both? If it should be possible to return multiple figures, they probably should be returned as a list of figures.

Tomorrow during the sprint, I’ll try to move further into this rabbit hole. Happy to chat then!

[bloyl]

yes that function is unusable for anything other than plotting a single TFR. If you look at the code the return is inside the loop so it shouldn’t be surprising that it doesn’t work with multiple channels. https://github.com/mne-tools/mne-python/blob/056c273f1bb96bb3229e60874122f830f946746c/mne/time_frequency/tfr.py#L1280-L1284

For what its worth it is relatively easy to write your own wrapper that will do what you need it to. I use something (https://gist.github.com/bloyl/a08a64899f2feffda5271890a5c6a677) based off the plotting in https://mne.tools/stable/auto_examples/time_frequency/plot_time_frequency_erds.html#sphx-glr-auto-examples-time-frequency-plot-time-frequency-erds-py to plot multi channel tfrs.