Closed haesleinhuepf closed 2 years ago
I wonder if you can get the correct behavior by not setting translucent
as your own default blending? Using None
instead? Then only set blending when you want additive or specifically depth checking, etc.
Then in 0.4.15 you will get the napari default: translucent
and in 0.4.16 you will get translucent_no_depth
The relevant code is here for single channel:
https://github.com/napari/napari/blob/23bac0cc9a8f2967934c255481936f8b2d809421/napari/components/viewer_model.py#L740
And here for split channel:
https://github.com/napari/napari/blob/23bac0cc9a8f2967934c255481936f8b2d809421/napari/layers/utils/stack_utils.py#L83
Edit: note that's the 0.4.16 code, so translucent
has been changed to translucent_no_depth
However, we would need to pin napari>=0.4.16 to make it work and 0.4.16 is not on conda yet.
you may check the napari version and provide a default value based on this.
Thinking more, I can't think of any issue with using translucent_no_depth
in 0.4.15, so if you do want to specify your own default, I think that should work?
I wonder if you can get the correct behavior by not setting
translucent
as your own default blending? UsingNone
instead?
I checked making this change locally and it works:
In 0.4.16 blending is translucent_no_depth
so the depth checking issue is resolved.
In 0.4.15 blending is translucent
and no change in behavior is observed.
(using the workflow in the original napari issue from the OP)
Not sure if it breaks something else, but it shouldn't? 🤣
Hey guys @Czaki @psobolewskiPhD ,
if you want to give it a try to pass None
or translucent_no_depth
and see if it works with both napari version, your PR is very welcome! :-)
Thanks! Robert
@haesleinhuepf Since I started playing with it locally, I may as well make the PR.
But, I notice that some of the Categories also hard-set blending to translucent
:
e.g. Measure Labels:
https://github.com/haesleinhuepf/napari-assistant/blob/fe4a40342ff232acddaca863209dbf0e94e20e7f/napari_assistant/_categories.py#L153-L163
How should those be handled? Maybe that explicit setting of blending should be dropped, so also just use the Category class default (which will be None, which will then use the napari default)? Or is there specific need for depth checking there?
The default was 'translucent', thus, this line in specific categories should not have any effect. You can try removing those ...
Right, but with the new default None
to fix the issue, those lines will hard-set translucent
again, which will cause the glitch in the napari issue for those specific categories.
So I was wondering if the setting for these Categories was intentional/necessary.
I guess not, so I will try removing and we'll see.
those lines will hard-set
translucent
again
That's why I proposed:
You can try removing those ...
solved in #19
Instead of
translucent
we should usetranslucent_no_depth
here: https://github.com/haesleinhuepf/napari-assistant/blob/fe4a40342ff232acddaca863209dbf0e94e20e7f/napari_assistant/_categories.py#L30... as discussed in https://github.com/napari/napari/issues/4618
However, we would need to pin napari>=0.4.16 to make it work and 0.4.16 is not on conda yet.