Repository navigation
[ENH]: FancyArrow in 3D #22571
Description
Activity
@patquem Can u give some insight s on how to solve this issue
Adding Arrow3D class in mplot3d, this should be easy :) .
Solving the arrow head problem during visualization, I have no idea how to fix that. Sorry :( .Hello @soma2000-lang
introducing the penultimate line in the code (if not....) avoids a warning message and drawing the wrong arrowhead as mentioned above. Not sure if this is the right way to fix the bug, but temporarily it might be a solution.
Patrickclass Arrow3D(FancyArrowPatch): """ Create fancy arrow (artist) for 3d matplotlib visualisation """ def __init__(self, xs, ys, zs, *args, **kwargs): FancyArrowPatch.__init__(self, (0, 0), (0, 0), *args, **kwargs) self._verts3d = xs, ys, zs def draw(self, renderer): xs3d, ys3d, zs3d = self._verts3d xarr, yarr, _ = proj3d.proj_transform(xs3d, ys3d, zs3d, self.axes.M) self.set_positions((xarr[0], yarr[0]), (xarr[1], yarr[1])) if not (np.allclose(xarr, 0.) and np.allclose(yarr, 0.)): FancyArrowPatch.draw(self, renderer)Hi, is anyone working on this ? I know you don't assign issues but I would like to try it !
I believe that maybe @soma2000-lang is not actively working on this, so please give it a go!
I have labeled this a a good first issue because there is ~working code for the implementation, but medium because it involves (limited) API design.
This will need:
- a method added to Axes3D
- numpydoc style docstring
- tests
- get to a consensus on the API (which in addition to the actually time it takes to have the conversation may take very long in calendar time)
- a whats-new entry
- an example in the gallery
This is a good project for someone with library design experience but who is new to Matplotlib.
Reacted by patquem and Caitlin Dancer- a method added to Axes3D
So I'm just gonna throw out this sketch because the signature is closer to FancyArrowPatch
class FancyArrowPatch3D(mpatches.FancyArrowPatch): def __init__(self, posA, posB, *args, **kwargs): mpatches.FancyArrowPatch.__init__(self, (0, 0), (0, 0), *args, **kwargs) self._verts3d = list(zip(posA, posB))
Here is a function based on the solution from @tacaswell which takes an array of arrow start points and endpoints. Additionally, it sets the axes limits so that all start/end points remain visible.
def arrows3d(ends, starts=None, ax=None, label=None, **kwargs): """3D plot of multiple arrows Args: ends (ndarray): (N, 3) size array of arrow end coordinates starts (ndarray): (N, 3) size array of arrow start coordinates. Assume start position of (0, 0, 0) if not given ax (Axes3DSubplot): existing axes to add to label (str): legend label to apply to this group of arrows kwargs (dict): additional arrow properties """ if starts is None: starts = np.zeros_like(ends) assert starts.shape == ends.shape, "`starts` and `ends` shape must match" assert len(ends.shape) == 2 and ends.shape[1] == 3, \ "`starts` and `ends` must be shape (N, 3)" class Arrow3D(FancyArrowPatch): def __init__(self, xs, ys, zs, *args, **kwargs): super().__init__((0,0), (0,0), *args, **kwargs) self._verts3d = xs, ys, zs def do_3d_projection(self, renderer=None): xs3d, ys3d, zs3d = self._verts3d xs, ys, zs = proj3d.proj_transform(xs3d, ys3d, zs3d, self.axes.M) self.set_positions((xs[0],ys[0]),(xs[1],ys[1])) return np.min(zs) # create new axes if none given if ax is None: ax = plt.figure().add_subplot(111, projection='3d') arrow_prop_dict = dict(mutation_scale=20, arrowstyle='-|>', color='k', shrinkA=0, shrinkB=0) arrow_prop_dict.update(kwargs) for ind, (s, e) in enumerate(np.stack((starts, ends), axis=1)): a = Arrow3D( [s[0], e[0]], [s[1], e[1]], [s[2], e[2]], # only give label to first arrow label=label if ind == 0 else None, **arrow_prop_dict ) ax.add_artist(a) # store starts/ends on the axes for setting the limits ax.points = np.vstack((starts, ends, getattr(ax, 'points', np.empty((0, 3))))) ax.set_xlim3d(ax.points[:, 0].min(),ax.points[:, 0].max()) ax.set_ylim3d(ax.points[:, 1].min(),ax.points[:, 1].max()) ax.set_zlim3d(ax.points[:, 2].min(),ax.points[:, 2].max()) return ax
ax = arrows3d(np.random.normal(size=(5, 3)), label='Foo') ax = arrows3d(np.random.normal(size=(5, 3)), ax=ax, color='r',label='Bar') plt.legend() plt.show()
heyy @patquem looking to the issue , seems like it can be done.
Let me work on this issue and fix it.@ShubhamPatel-16 this issue can't be moved forward until decisions are made in #22435 b/c this API should match that one as much as possible.
Removing the good first issue label for now b/c this isn't ready for work yet.
Reacted by SHUBHAM PATEL and patquem
Problem
Hello,
FancyArrow can not be used in 3D.
Some proposals (#21688) consists in creating a new class.
Could it be possible:
Thanks,
Patrick
import numpy as np import matplotlib.pyplot as plt from matplotlib.patches import FancyArrowPatch, FancyArrow from mpl_toolkits.mplot3d import proj3d class Arrow3D(FancyArrowPatch): """ Create fancy arrow (artist) for 3d matplotlib visualisation """ def __init__(self, xs, ys, zs, *args, **kwargs): FancyArrowPatch.__init__(self, (0, 0), (0, 0), *args, **kwargs) self._verts3d = xs, ys, zs def draw(self, renderer): xs3d, ys3d, zs3d = self._verts3d xarr, yarr, _ = proj3d.proj_transform(xs3d, ys3d, zs3d, self.axes.M) self.set_positions((xarr[0], yarr[0]), (xarr[1], yarr[1])) FancyArrowPatch.draw(self, renderer) fig, ax = plt.subplots(1, 2, figsize=(12, 5), subplot_kw=dict(projection="3d", proj_type='ortho')) ax[0].voxels(np.ones((1, 1, 1)), facecolors=[0, 0, 0, 0], edgecolors='k') # ax[0].add_artist(FancyArrow(0, 0, 0, 1, 1, 1)) ax[0].add_artist(Arrow3D((0, 1), (0, 1), (0, 1), mutation_scale=50)) ax[1].voxels(np.ones((1, 1, 1)), facecolors=[0, 0, 0, 0], edgecolors='k') ax[1].view_init(elev=28, azim=45) # ax[1].add_artist(FancyArrow(0, 0, 0, 1, 1, 1))) ax[1].add_artist(Arrow3D((0, 1), (0, 1), (0, 1), mutation_scale=50)) plt.show()Proposed solution
No response