Selecting multiple objects

Here we show how to select objects in the 3D world. In this example all objects to be picked are part of a single actor.

FURY likes to bundle objects in a few actors to reduce code and increase speed. Nonetheless the method works for multiple actors too.

The difference with the picking tutorial is that here we will be able to select more than one object. In addition we can select interactively many vertices or faces.

In summary, we will create an actor with thousands of cubes and then interactively we will be moving a rectangular box by hovering the mouse and making transparent everything that is behind that box.

import numpy as np
from fury import actor, window, utils, pick

Adding many cubes of different sizes and colors

num_cubes = 50000

centers = 10000 * (np.random.rand(num_cubes, 3) - 0.5)
colors = np.random.rand(num_cubes, 4)
colors[:, 3] = 1.0
radii = 100 * np.random.rand(num_cubes) + 0.1

Keep track of total number of triangle faces Note that every quad of each cube has 2 triangles and each cube has 6 quads in total.

num_faces = num_cubes * 6 * 2

Build scene and add an actor with many objects.

Build the actor containing all the cubes

cube_actor = actor.cube(centers, directions=(1, 0, 0),
                        colors=colors, scales=radii)

Access the memory of the vertices of all the cubes

Access the memory of the colors of all the cubes

vcolors = utils.colors_from_actor(cube_actor, 'colors')

Create a rectangular 2d box as a texture

rgba = 255 * np.ones((100, 200, 4))
rgba[1:-1, 1:-1] = np.zeros((98, 198, 4)) + 100
texa = actor.texture_2d(rgba.astype(np.uint8))


Create the Selection Manager

selm = pick.SelectionManager(select='faces')

Tell Selection Manager to avoid selecting specific actors


Let’s make the callback which will be called when we hover the mouse

def hover_callback(_obj, _event):
    event_pos = selm.event_position(showm.iren)
    # updates rectangular box around mouse
    texa.SetPosition(event_pos[0] - 200//2,
                     event_pos[1] - 100//2)

    # defines selection region and returns information from selected objects
    info =, showm.scene, (200//2, 100//2))
    for node in info.keys():
        if info[node]['face'] is not None:
            if info[node]['actor'] is cube_actor:
                for face_index in info[node]['face']:
                    # generates an object_index to help with coloring
                    # by dividing by the number of faces of each cube (6 * 2)
                    object_index = face_index // 12
                    sec = int(num_vertices / num_objects)
                    color_change = np.array([150, 0, 0, 255], dtype='uint8')
                    vcolors[object_index * sec: object_index * sec + sec] \
                        = color_change

Make the window appear

showm = window.ShowManager(scene, size=(1024, 768),

Bind the callback to the actor


Change interactive to True to start interacting with the scene

interactive = False

if interactive:


Save the current framebuffer in a PNG file

window.record(showm.scene, size=(1024, 768), out_path="viz_selection.png")

Total running time of the script: ( 0 minutes 2.922 seconds)

Gallery generated by Sphinx-Gallery