Set animation speed (FPS) for Open3D

[timwoelfle]

Thank you very much for this awesome package, I'm using it to visualize smartphone IMU data and like it a lot! While I've found how to set the animation speed in matplotlib (interval argument, see here), I haven't found any way to set it for Open3D animate. Have I missed something? Otherwise it would be great if this could be integrated so that my animations frames per seconds are on par with the IMU sensor reading frequency!

[AlexanderFabisch]

Thank you very much for this awesome package, I'm using it to visualize smartphone IMU data and like it a lot!

Thanks. :)

I'm using it to visualize smartphone IMU data and like it a lot! While I've found how to set the animation speed in matplotlib (interval argument, see here), I haven't found any way to set it for Open3D animate. Have I missed something? Otherwise it would be great if this could be integrated so that my animations frames per seconds are on par with the IMU sensor reading frequency!

I think Open3D doesn't directly support this, so this is something that would have to be done in pytransform3d.

As a workaround, I would suggest to modify your animation callback to sleep for the correct amount of time. If you expect that your animation does almost need no time you can just call import time; time.sleep(seconds as float). Otherwise you can make a callable object that keeps track of the time that you have to sleep in between.

[timwoelfle]

Thank you very much for your quick reply! Alternatively I could try the whole thing with matplotlib where I can already set FPS with the interval argument, but I'm not sure how to transform a box in a matplotlib animation (I did manage in Open3D)? Unfortunately, this does not work (rather redraws many basis-frames and doesn't transform the box itself during the animation):

def update_box(step, n_frames, ax):
    A2B = A2B_from_frame(step)
    plot_transform(ax=ax, A2B=A2B)
    return ax

fig = plt.figure()

ax = make_3d_axis(ax_s=1)

box = plot_box(ax=ax, size=[0.5, 1, 0.05], A2B=A2B, wireframe=True, alpha=0.3, color="k")
plot_transform(ax=ax, A2B=A2B)

anim = animation.FuncAnimation(
       fig, update_box, n_frames, fargs=(n_frames, ax),
       interval=20, blit=False)

plt.show()

I've successfully animated a Frame by modifying your example, but I'm not sure how to add the box:

def update_frame(step, n_frames, frame):
    A2B = A2B_from_frame(step)
    frame.set_data(A2B)
    return frame

fig = plt.figure()

ax = make_3d_axis(ax_s=1)

frame = Frame(A2B)
frame.add_frame(ax)

anim = animation.FuncAnimation(
       fig, update_frame, n_frames, fargs=(n_frames, frame),
       interval=20, blit=False)

plt.show()

(Sorry, I know this doesn't exactly belong here, but maybe you're able to help :-) )

[AlexanderFabisch]

Matplotlib is not the best tool for 3d animations. I'd recommend some real 3d visualization software for more complex cases.

As you can see here, you need to modify the Matplotlib (artist?) objects in each frame. These will be created here. plot_box will not return them. I'd recommend to clone the code from plot_box if you want to use matplotlib for these animations.

Unforunately, I don't have a good recommendation as an alternative to Matplotlib or Open3D in Python, but you could take a look at pybullet (which is mainly a physics engine, but can also be used for simple real-time rendering). Take a look at the functions

• pybullet.connect(pybullet.GUI) to start the GUI
• pybullet.createVisualShape and pybullet.createMultiBody to create a box
• pybullet.resetBasePositionAndOrientation to set position and orientation (quaternion, scalar last!)
• pybullet.setRealTimeSimulation to render in real time, example: https://github.com/bulletphysics/bullet3/blob/21d1b8fc7183eed5903cf7bfe9ef8a81e432ab15/examples/pybullet/gym/pybullet_examples/biped2d_pybullet.py#L33

[timwoelfle]

Thanks a lot, I'll check it out!