Wrong colormap for H&E

[marcovarrone]

Hello, maybe I am doing something wrong because I thought this problem would not occur after #127.
When I follow the tutorial on the Visium dataset and I plot the H&E image with the code shown in the tutorial I get the following image:
.

I checked and my version of spatialdata_plot contains the edit done in that pull request.
Cheers!

[timtreis]

Hey @marcovarrone, which plot would you expect? I think we haven't updated the tutorial notebook after #127, so that could explain the difference?

[marcovarrone]

If you plot the image using matplotlib, it shows the sample in the way the H&E usually looks like.

plt.imshow(np.transpose(np.array(sdata.images["ST8059048_image"]), axes=(1,2,0)))

So this:

[timtreis]

Oh weird 🤔 My go-to check for this was always our raccoon picture which now still has the correct colors. I'll try to figure out why our H&E looks different

[marcovarrone]

Thank you very much for looking into it!
I also tried with my own H&E image and I got the same problem.

[LucaMarconato]

It could be maybe to an alpha channel present in the H&E image that makes the plot escapes some if conditions? @marcovarrone could you please try with a rgb version of the H&E image, if it is a rgba image?

[giovp]

hitting the same issue, @timtreis could you please take a look, it's rather weird as it should just use plt.imshow int he backed and I'm fairly confident that at first release this was working correctly so it must have been something introduces later on.

@marcovarrone you can try something like this for the time being

gene = "TAS1R1"
fig, ax = plt.subplots(1,1,figsize=(10,10))
ax.imshow(sdatas["ht232p1h1a2"].images["ht232p1h1a2_image"].transpose("y","x","c"))
sdatas["ht232p1h1a2"].pl.render_shapes(color=gene, layer="X").pl.show(ax=ax)

[timtreis]

Found that it was caused by a default normalisation. Going to disable this with a PR later