FIX: size and color rendering for Path3DCollection by tacaswell · Pull Request #19812 · matplotlib/matplotlib · GitHub
Skip to content

FIX: size and color rendering for Path3DCollection - #19812

Merged
QuLogic merged 1 commit into
matplotlib:masterfrom
tacaswell:fix_scatter3D
Mar 31, 2021
Merged

FIX: size and color rendering for Path3DCollection#19812
QuLogic merged 1 commit into
matplotlib:masterfrom
tacaswell:fix_scatter3D

Conversation

@tacaswell

@tacaswell tacaswell commented Mar 29, 2021

Copy link
Copy Markdown
Member

PR Summary

This fixes both bugs identified in #19787

Still needs test and the tactic should be expanded to the other *Collections before we merge.

@tacaswell tacaswell added the Release critical For bugs that make the library unusable (segfaults, incorrect plots, etc) and major regressions. label Mar 29, 2021
@tacaswell tacaswell added this to the v3.4.1 milestone Mar 29, 2021
@tacaswell tacaswell mentioned this pull request Mar 29, 2021
Closed
QuLogic
QuLogic reviewed Mar 30, 2021
View reviewed changes
Comment thread lib/mpl_toolkits/mplot3d/art3d.py
@tacaswell
tacaswell marked this pull request as ready for review March 30, 2021 22:42
@jklymak

jklymak commented Mar 31, 2021

Copy link
Copy Markdown
Member

@tacaswell

tacaswell commented Mar 31, 2021

Copy link
Copy Markdown
Member Author

I was thinking about the fix @QuLogic put in for #18916 and was wondering if that was going to come back in other ways and it did!

@tacaswell

tacaswell commented Mar 31, 2021

Copy link
Copy Markdown
Member Author

No, I jumped to conclusion, but it is an easy fix. Running a full docs build locally before pushing.

@QuLogic

QuLogic commented Mar 31, 2021

Copy link
Copy Markdown
Member

I think you're running into the same issue as #18932 did, namely that legend copies the class, and so creates a 3D collection for its legend handle, without doing any of the 3D work.

@tacaswell

tacaswell commented Mar 31, 2021

Copy link
Copy Markdown
Member Author

I think this is slightly different, the problem is it is actually creating a the 3D class, rather than the slightly-upsetting

matplotlib/lib/mpl_toolkits/mplot3d/art3d.py

Lines 705 to 728 in bba76b8

def patch_collection_2d_to_3d(col, zs=0, zdir='z', depthshade=True):
"""
Convert a :class:`~matplotlib.collections.PatchCollection` into a
:class:`Patch3DCollection` object
(or a :class:`~matplotlib.collections.PathCollection` into a
:class:`Path3DCollection` object).
Parameters
----------
za
The location or locations to place the patches in the collection along
the *zdir* axis. Default: 0.
zdir
The axis in which to place the patches. Default: "z".
depthshade
Whether to shade the patches to give a sense of depth. Default: *True*.
"""
if isinstance(col, PathCollection):
col.__class__ = Path3DCollection
elif isinstance(col, PatchCollection):
col.__class__ = Patch3DCollection
col._depthshade = depthshade
col.set_3d_properties(zs, zdir)
so in the case of scatter we were creating it via the 3d props call, but in the scatter case we were running the __init__ which bounced to the super class which tries to set the linewidth before setup the 3D stuff.

@tacaswell

tacaswell commented Mar 31, 2021

Copy link
Copy Markdown
Member Author

QuLogic
QuLogic approved these changes Mar 31, 2021
View reviewed changes

@QuLogic QuLogic left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have minor comments, and I'll leave this open for a little bit in case anyone else is able to review.

Comment thread lib/mpl_toolkits/mplot3d/art3d.py Outdated
Comment thread lib/mpl_toolkits/mplot3d/art3d.py
Comment thread lib/mpl_toolkits/mplot3d/art3d.py Outdated
Comment thread lib/mpl_toolkits/tests/test_mplot3d.py Outdated
QuLogic
QuLogic reviewed Mar 31, 2021
View reviewed changes
Comment thread lib/mpl_toolkits/mplot3d/art3d.py
Comment thread lib/mpl_toolkits/mplot3d/art3d.py Outdated
@tacaswell @QuLogic
FIX: size and color rendering for Path3DCollection
f888285 
This fixes matplotlib#19787

This modifies the way we manage internal depth sorting of elements in the 3D
collections.  We need to both preserve the original order (so we can keep the
position, size, colour, ... in sync), but present each of them to the 2D
rendering code in depth order (back to front).  In some cases we can handle
this by sorting the values in the getter methods and in others we have to keep
a cache of both the sorted (the names used by the 2D code) and
unsorted (post-fixed with 3d) data.

It does not appear that Line3DCollection does any of the z-sorting that the
other collections do.  While definitely a limitation and produces MC Escher
effects, this is out of scope of the current work.

The z-sorting on Poly3DCollection is significantly more sophisticated than the
sorting on Path3DCollection and Patch3DCollection so it is handled
differently.

Co-authored-by: Elliott Sales de Andrade <quantum.analyst@gmail.com>
@QuLogic
QuLogic merged commit ee9500a into matplotlib:master Mar 31, 2021
QuLogic added a commit to QuLogic/matplotlib that referenced this pull request Mar 31, 2021
@QuLogic
Backport PR matplotlib#19812: FIX: size and color rendering for Path3…
040649d 
…DCollection
QuLogic added a commit that referenced this pull request Mar 31, 2021
@QuLogic
Merge pull request #19834 from QuLogic/backport-19812
17b3aa0 
Backport PR #19812: FIX: size and color rendering for Path3DCollection
@tacaswell
tacaswell deleted the fix_scatter3D branch March 31, 2021 13:36
@tacaswell tacaswell mentioned this pull request Apr 2, 2021
Open
@oscargus oscargus mentioned this pull request Jul 22, 2022
Merged
2 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@QuLogic QuLogic QuLogic approved these changes

Assignees

No one assigned

Labels

Release critical For bugs that make the library unusable (segfaults, incorrect plots, etc) and major regressions.

Projects

None yet

Milestone

v3.4.1

Development

Successfully merging this pull request may close these issues.

3 participants

@tacaswell @jklymak @QuLogic