Jérôme Dockès

> I'm quite uncertain whether we should do it (these artlases are convenient for users Fair enough! It's just that as @NicolasGensollen worked on #3083 we realized their interface is...

Here is what was decided in the office hours. for the NiftiLabelsMasker: - PR 1: - improve the documentation of the NiftiLabelsMasker's `labels` parameter to make it clear that `labels[j]`...

thanks for looking into this! - the zip download did not work for large collections last time I checked (~ 1 year ago), whether through the API or the website...

Hi, there may be a difference of 1/256th of the length of the colorbar because the colormap is discrete https://github.com/nilearn/nilearn/blob/d43e4ff0e150634c474f834d523ada1529d218d1/nilearn/plotting/surf_plotting.py#L1109), or it might just be the thickness of the tick...

thanks! having the image is useful to reproduce the problem. so to summarize, the cmap could be made a bit more symmetric by taking the next integer here rather than...

Thanks a lot @alexisthual I can't do a full review this week but I think the new plots are really cool. Actually I think we should make it easy to...

> Maybe we could use `fsaverage7` in the first image so that it looks nicer. For now, it looks like this on `fsaverage5` (with matplotlib and plotly respectively): that's true...

> Otherwise, I noticed that the 'darkness' parameter is reminicent of the 'dim' parameter in plot_img functions. We should probably reunify the API here, but go first through deprecation cycles....

I understand the motivation for allowing to choose a different background in `view_img_on_surf` but it should be replaced with a sensible default. ATM the default is sulcal depth of fsaverage...

> Actually, I did a few more tests: `darkness=0` gives a perfectly white background, while `darkness=1` allows to display dark shades, so maybe renaming it to `bg_brightness` doesn't make sense...