why `render_channels` is a separate method and it's not an argument in the `render_image` method? about spatialdata-plot HOT 2 CLOSED

Yes good point, Tim and I already discussed that. The render_channels method currently handles (unnormalised) multichannel images (usually with more than 3 channels). This is useful for CODEX/IMC type image data in which you often have 50+ markers/channels. Here you are often interested in visualising 1-5 channels with different colors and thus you'd need to first select the appropriate channels and assign then colors to them.

One question concerns at which point we want the user to select channels. Currently, this is done in pl.render_channels, e.g.

sdata.pl.render_channels(channels=[1,2,3], colors=['blue', 'red', 'green']).pl.show()

but perhaps more consistent with our .pp preprocessing logic would be to enforce something like this

sdata.pp.get_elements('image1', channels=[1,2,3]).pl.render_channels(['blue', 'red', 'green']).pl.show()

or even

sdata.pp.get_elements('image1').pp.get_channels([1,2,3]).pl.render_channels(['blue', 'red', 'green']).pl.show()

Do we have a preference regarding that ?

Generally, for the sake of simple API, I am also in favour of refactoring the pl.render_channels into pl.render_images.

from spatialdata-plot.

I think render_images should have channels and cmap arguments anyway so the channel selection should just be part of render_images

from spatialdata-plot.