FIX: colormapping of integer normed images #6581
Conversation
|
Tried to make a PR against this, but we have too many contributors... ? ;) Check out my branch for an additional fix for boundary norm: https://github.com/mdboom/matplotlib/tree/image-over-under |
|
You have to just edit the URL. They are aware of the problem, conclude it affects about 15 projects and oh well. |
|
@tacaswell, https://github.com/efiring/matplotlib/tree/tacaswell-fix_over_under_images has my proposed fix on top of yours. I don't quite understand the procedure for handling such combinations of commits; maybe it is simplest at this point for you to pull my commit from my tree and add it to this PR. I haven't done any testing other than verifying that it makes 'image_masked.py' work. |
Use the same trick for carrying through the masked points to use the green and blue channels to carry the under/over markers.
tacaswell
force-pushed
the
fix_over_under_images
branch
from
c1ec9fa to
1ec7071
Compare
For the next person who has to read this code.
tacaswell
changed the title
| elif A.ndim == 3: | ||
| # colormap norms that output integers (ex NoNorm | ||
| # and BoundaryNorm) to RGBA space before | ||
| # interpolating. This is needed due to the the |
|
Tests would be nice. At least an image comparison test based on the example in #6509 seems like a good idea |
|
I'll add tests + fix the text tonight. |
|
I've said this "in person", but for the written record -- this seems like the right approach to me. I'm 👍 on merging this, modulo everyone else's comments above. |
This is a work in progress and will hopefully be a starting point for @mdboom to fix it properly 😄
Starts to address #6509