matplotlib: getting coordinates in 3D plots by a mouseevent

Question

I want to get coordinates (x,y,z) in 3D plots by a mouse event such as a click. MATLAB has this function, datacursormode. A good image is in the following link.

http://www.mathworks.com/help/matlab/ref/datacursormode.html

mpldatacursor (https://github.com/joferkington/mpldatacursor) is a similar function for matplotlib, however, this seems to be unsuitable for 3D plots. x and y values are not proper even though they can be get.

from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as np
from mpldatacursor import datacursor

x = np.arange(-3, 3, 0.25)
y = np.arange(-3, 3, 0.25)
X, Y = np.meshgrid(x, y)
Z = np.sin(X)+ np.cos(Y)

fig = plt.figure()
ax = Axes3D(fig)
surf = ax.plot_surface(X, Y, Z, rstride=1, cstride=1)

datacursor(surf)

plt.show()

I also want to get z value, if it is possible. Is there any good way?

Surprising that this seems unsupported. – MattCochrane Jun 25 '16 at 07:29 — MattCochrane, Jun 25 '16 at 07:29

score 0 · Answer 1 · edited May 23 '17 at 12:18

0

According to the file "changelog.rst" at the link you suggested (https://github.com/joferkington/mpldatacursor) this function has been added in July 2015. Unfortunately it looks like it extracts the data points from the location where the mouse clicks rather than the original data set. This leads to some imprecision in the result. A possibility could be to modify the datacursor according to the instructions provided for the 2D version in Get data from plot with matplotlib. Hope this helps.

edited May 23 '17 at 12:18

Community

1
1

answered May 11 '17 at 13:53

Giovanni.R88

129
1
6

While it's true the cursor location is used instead of the nearest point in the dataset, the error also comes from the fact that a cursor location is known in 2D, and so refers to an infinite number of possible 3D points, any point along the line under the cursor (which depends on the projection). *matplotlib* implements a rule to decide which point it shows (the same rule *mpldatacursor* uses). In the general case, it won't yield a point in the dataset. Continued... – Jim Oct 17 '20 at 20:44
...To get that, you'd have to implement some other rule, like finding the point in the dataset nearest to the line under the cursor. To make that search efficient for lots of points, storing them in a hierarchical data structure can help. – Jim Oct 17 '20 at 20:47

score -1 · Answer 2 · answered Jul 25 '18 at 15:17

mpldatacursor was too complicated for what I wanted, which was to receive the x,y,z coordinates in a callback function somehow. I extracted a helper function (get_xyz_mouse_click) from mpldatacursor pick_info.py that does the bare minimum needed to get the coordinates (i.e., no hover window, no complicated event processing). Here's the helper function:

import numpy as np
import matplotlib.transforms as mtransforms
from mpl_toolkits import mplot3d

def get_xyz_mouse_click(event, ax):
    """
    Get coordinates clicked by user
    """
    if ax.M is None:
        return {}

    xd, yd = event.xdata, event.ydata
    p = (xd, yd)
    edges = ax.tunit_edges()
    ldists = [(mplot3d.proj3d.line2d_seg_dist(p0, p1, p), i) for \
                i, (p0, p1) in enumerate(edges)]
    ldists.sort()

    # nearest edge
    edgei = ldists[0][1]

    p0, p1 = edges[edgei]

    # scale the z value to match
    x0, y0, z0 = p0
    x1, y1, z1 = p1
    d0 = np.hypot(x0-xd, y0-yd)
    d1 = np.hypot(x1-xd, y1-yd)
    dt = d0+d1
    z = d1/dt * z0 + d0/dt * z1

    x, y, z = mplot3d.proj3d.inv_transform(xd, yd, z, ax.M)
    return x, y, z

Here's an example of using it in a 3D scatter plot:

from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as np

import simple_pick_info.pick_info

# Fixing random state for reproducibility
np.random.seed(19680801)


def randrange(n, vmin, vmax):
    '''
    Helper function to make an array of random numbers having shape (n, )
    with each number distributed Uniform(vmin, vmax).
    '''
    return (vmax - vmin)*np.random.rand(n) + vmin

fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')

n = 100

# For each set of style and range settings, plot n random points in the box
# defined by x in [23, 32], y in [0, 100], z in [zlow, zhigh].
for c, m, zlow, zhigh in [('r', 'o', -50, -25), ('b', '^', -30, -5)]:
    xs = randrange(n, 23, 32)
    ys = randrange(n, 0, 100)
    zs = randrange(n, zlow, zhigh)
    ax.scatter(xs, ys, zs, c=c, marker=m)

ax.set_xlabel('X Label')
ax.set_ylabel('Y Label')
ax.set_zlabel('Z Label')

plt.show()


def on_press(event):
    x,y,z = simple_pick_info.pick_info.get_xyz_mouse_click(event, ax)    
    print(f'Clicked at: x={x}, y={y}, z={z}')

cid = fig.canvas.mpl_connect('button_press_event', on_press)

You should edit on_press() to do something with x,y,z. It still has the problem that the other answer cited with using the axes grid to produce a point (i.e., it does not search the original data for a nearest neighbor). I recommend doing that with a distance transform to your original data model (points, lines, etc.) because it will get very complicated to search the patches in a surface.

I really wish this was built into matplotlib in the way that Matlab's datacursormode works!

I get the error `ModuleNotFoundError: No module named 'simple_pick_info'`. This module does not seem to exist. — Adriaan, Mar 27 '19 at 13:44
the solution doesnt make it clear that the helper function would usually reside in the module simple_pick_info . in the example, just change x,y,z = simple_pick_info.pick_info.get_xyz_mouse_click(event, ax) to x,y,z = get_xyz_mouse_click(event, ax) it doesnt provide mouseover, but does output the mouse coordinate position. — lee, Jan 24 '22 at 19:42

matplotlib: getting coordinates in 3D plots by a mouseevent

2 Answers2