Skip to content

Conversation

psobolewskiPhD
Copy link

I mentioned this here:
#8 (comment)

For the images I tested and even the Cells 3D example, the 0.3 opacity was too low for me to really see that the segmentation worked!
In this PR I bump it to 0.6 which to me looks pretty good with the new aggregation too.
(tested locally)

@psobolewskiPhD psobolewskiPhD changed the title bump opacity of labels layers to 0.6 Minor end: bump up the opacity of labels layers to 0.6 Mar 24, 2025
@psobolewskiPhD psobolewskiPhD changed the title Minor end: bump up the opacity of labels layers to 0.6 Minor ENH: bump up the opacity of labels layers to 0.6 Mar 24, 2025
@FabianIsensee
Copy link
Member

If you modify the opacity of the nnInteractive label layer this change should stick for the entire session. I think this is a more reasonable approach. Other segmentation problems require the lower opacity on order to properly see the objects border while iteratively refining the segmentation. We initially had a higher opacity and lowered it because we noticed people making errors because they could no longer see the image well enough.
What are your thoughts @Lars-Kraemer ?

@psobolewskiPhD
Copy link
Author

That totally makes sense. It might be hard to find one universal value.
My issue was that I had a hard time telling if the segmentation worked, because the labels were so subtle.
Totally fine with y'all closing this. Thanks for the awesome tool!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants