Partial topomaps (masking?)

[mcaffini]

In my effort to use MNE Python with fNIRS data I find that a major issue is to visualize topomaps. In most fNIRS systems, channels come in limited number and only cover part of the scalp, while in EEG and MEG the whole scalp is usually covered.
It would be usuful to have some sort of masking support in plot_topomap, for example using a convex hull approach (quoting @agramfort). I'm not sure whether this can be achieved using other parameters such as outlines and image_mask and, if so, how.

p.s. Probably a new name for this feature should be introduced, because the parameter name 'mask' is taken (by mask and mask_params) and is used for a different purpose.

[agramfort]

could you share a small -ave.fif file to test?

[mcaffini]

Sorry I had to wait for a vacation day to get back to this.
This is an evoked file with my montage. In the layout plot it does not render properly because it normalizes the coordinates though.

evoked-ave.zip

[agramfort]

thanks for sharing we'll look into it

[larsoner]

Closing this as a dup of #3987, @mcaffini feel free to subscribe to that issue for updates