.ipynb .pdf

Contents

pyrecorder

Github: https://github.com/anyoptimization/pyrecorder

Installation

The framework is available at the PyPi Repository:

pip install -U pyrecorder

Matplotlib

Please note that the example below are using the vp80 codec to create a video which can be played in a browser and also this documentation. Nevertheless, without specifing the codec mp4v is used by default.

Video

import numpy as np
import matplotlib.pyplot as plt

from pyrecorder.recorder import Recorder
from pyrecorder.writers.video import Video
from pyrecorder.converters.matplotlib import Matplotlib

# create a writer that takes the
writer = Video("example.webm", codec='vp80')

# use the with statement to close the recorder when done
with Recorder(writer) as rec:

    # use black background for this plot
    plt.style.use('dark_background')

    # record 10 different snapshots
    for t in range(50, 500, 5):

        a = np.arange(t) * 0.1
        plt.plot(a * np.sin(a), a * np.cos(a))
        plt.xlim(-50, 50)
        plt.ylim(-50, 50)
        plt.axis('off')

        # use the record to store the current plot
        rec.record()

# revert to default settings for other plots
plt.style.use('default')

When the code has finished, the video has been written to the specified filename example.mp4. Let us look what has been recorded:

display("example.webm")

For this example the default settings have been used and the global drawing space of Matplotlib is recorded. Let us look at another example with a few modifications:

import numpy as np
import matplotlib.pyplot as plt

from pyrecorder.recorder import Recorder
from pyrecorder.writers.video import Video
from pyrecorder.converters.matplotlib import Matplotlib

# initialize the converter which is creates an image when `record()` is called
converter = Matplotlib(dpi=120)

writer = Video("example2.webm", codec='vp80')

rec = Recorder(writer, converter=converter)

for t in range(10):

    # let us create a local figure object with two sub figures
    fig, (ax1, ax2) = plt.subplots(2, figsize=(3, 4))

    X = np.random.random((100, 2))
    ax1.scatter(X[:, 0], X[:, 1], color="green")

    X = np.random.random((100, 2))
    ax2.scatter(X[:, 0], X[:, 1], color="red")

    # fix the size of figure and legends
    fig.tight_layout()

    # take a snapshot the specific figure object with the recorder
    rec.record(fig=fig)


rec.close()

display("example2.webm")

GIF

import matplotlib.pyplot as plt
import numpy as np

from pyrecorder.recorder import Recorder
from pyrecorder.writers.gif import GIF

with Recorder(GIF("example.gif", duration=0.2)) as rec:

    for t in range(0, 200, 5):

        x = np.linspace(0, 4, 100)
        y = np.sin(2 * np.pi * (x - 0.01 * t))
        plt.plot(x, y)

        rec.record()

Contact

Feel free to contact me if you have any question:

Julian Blank (blankjul [at] egr.msu.edu)

Michigan State University

Computational Optimization and Innovation Laboratory (COIN)

East Lansing, MI 48824, USA