Blitting for performance in Matplotlib

Introduction

Blitting helps redraw only parts of a plot that change. This makes animations or updates faster and smoother.

When you want to animate a plot smoothly without redrawing everything.

When updating only a small part of a graph repeatedly, like a moving point.

When working with large plots and want to save computer resources.

When creating interactive visualizations that change often.

When you want to improve performance in real-time data displays.

Syntax

Matplotlib

fig, ax = plt.subplots()
background = fig.canvas.copy_from_bbox(ax.bbox)
# Draw static elements here
fig.canvas.blit(ax.bbox)
# Update dynamic elements
fig.canvas.restore_region(background)
# Draw updated elements
ax.draw_artist(dynamic_element)
fig.canvas.blit(ax.bbox)
fig.canvas.flush_events()

copy_from_bbox saves the background image of the plot area.

blit redraws only the changed parts for better speed.

Examples

This example moves a red dot diagonally by updating only the dot's position.

Matplotlib

import matplotlib.pyplot as plt
import numpy as np

fig, ax = plt.subplots()
line, = ax.plot([], [], 'ro')
ax.set_xlim(0, 10)
ax.set_ylim(0, 10)

background = fig.canvas.copy_from_bbox(ax.bbox)

for x in range(10):
    fig.canvas.restore_region(background)
    line.set_data(x, x)
    ax.draw_artist(line)
    fig.canvas.blit(ax.bbox)
    fig.canvas.flush_events()

Here, the static plot is drawn once and saved as background for future updates.

Matplotlib

fig, ax = plt.subplots()
ax.plot([1, 2, 3], [4, 5, 6])
background = fig.canvas.copy_from_bbox(ax.bbox)
# Static plot drawn once
fig.canvas.blit(ax.bbox)

Sample Program

This program animates a sine wave moving horizontally. It uses blitting to update only the sine wave line, making the animation smooth and fast.

Matplotlib

import matplotlib.pyplot as plt
import numpy as np
import time

plt.ion()  # Turn on interactive mode
fig, ax = plt.subplots()
ax.set_xlim(0, 2 * np.pi)
ax.set_ylim(-1.5, 1.5)
line, = ax.plot([], [], 'b-')

x = np.linspace(0, 2 * np.pi, 100)

# Draw static elements
fig.canvas.draw()
background = fig.canvas.copy_from_bbox(ax.bbox)

for phase in np.linspace(0, 2 * np.pi, 60):
    fig.canvas.restore_region(background)
    y = np.sin(x + phase)
    line.set_data(x, y)
    ax.draw_artist(line)
    fig.canvas.blit(ax.bbox)
    fig.canvas.flush_events()
    time.sleep(0.05)

plt.ioff()
plt.show()

Important Notes

Blitting works best with simple animations where only small parts change.

Interactive mode (plt.ion()) helps to see updates live.

Not all backends support blitting; use one that does (like TkAgg).

Summary

Blitting redraws only changed parts of a plot for better speed.

It is useful for animations and interactive updates.

Use copy_from_bbox, restore_region, and blit methods together.

Practice

1. What is the main purpose of blitting in matplotlib?

easy

A. To redraw only the changed parts of a plot for faster updates

B. To create 3D plots from 2D data

C. To save plots as image files

D. To change the color scheme of a plot

Solution

1. Step 1: Understand what blitting does

   Blitting redraws only the parts of the plot that change, instead of the whole plot.

2. Step 2: Compare options

   Options B, C, and D describe unrelated tasks like 3D plotting, saving files, or color changes.

3. Final Answer:

   To redraw only the changed parts of a plot for faster updates -> Option A

4. Quick Check:

   Blitting = redraw changed parts only [OK]

Hint: Blitting means updating only what changes fast [OK]

Common Mistakes:

• Thinking blitting saves plots as files
• Confusing blitting with changing colors
• Assuming blitting creates 3D plots

2. Which of the following is the correct way to save the background region for blitting in matplotlib?

easy

A. background = ax.copy_from_bbox(fig.bbox)

B. background = fig.canvas.copy_from_bbox(ax.bbox)

C. background = ax.copy_from_bbox(ax.bbox)

D. background = fig.copy_from_bbox(ax.bbox)

Solution

1. Step 1: Identify correct method usage

   The copy_from_bbox method is called on the figure canvas (fig.canvas) with the axes bounding box (ax.bbox).

2. Step 2: Check options carefully

   background = fig.canvas.copy_from_bbox(ax.bbox) is correct. Options B, C call it on ax (which lacks the method), D calls it on fig (missing .canvas), B also uses wrong bbox.

3. Final Answer:

   background = fig.canvas.copy_from_bbox(ax.bbox) -> Option B

4. Quick Check:

   copy_from_bbox called on fig.canvas with ax.bbox [OK]

Hint: copy_from_bbox called on fig.canvas with ax.bbox [OK]

Common Mistakes:

• Calling copy_from_bbox on ax instead of fig.canvas
• Using fig.bbox instead of ax.bbox
• Mixing up ax and fig in method calls

3. What will the following code print?

import matplotlib.pyplot as plt
fig, ax = plt.subplots()
line, = ax.plot([0, 1], [0, 1])
background = fig.canvas.copy_from_bbox(ax.bbox)
line.set_ydata([1, 0])
fig.canvas.restore_region(background)
ax.draw_artist(line)
print(line.get_ydata())

medium

A. [1 0]

B. [0 1]

C. [0 0]

D. Error: restore_region not found

Solution

1. Step 1: Trace the code changes

   The line's y-data is changed to [1, 0] using set_ydata.

2. Step 2: Understand blitting steps

   The background is restored, then the updated line is drawn. The line's data remains [1, 0].

3. Final Answer:

   [1 0] -> Option A

4. Quick Check:

   set_ydata changes line data to [1 0] [OK]

Hint: set_ydata changes data; restore_region redraws background [OK]

Common Mistakes:

• Assuming restore_region resets line data
• Confusing line data with original plot data
• Expecting an error from restore_region

4. You try to use blitting but your plot does not update visually after calling draw_artist. What is the most likely mistake?

medium

A. You called copy_from_bbox after draw_artist

B. You did not call plt.show() at the end

C. You used restore_region before copy_from_bbox

D. You forgot to call fig.canvas.blit(ax.bbox) after draw_artist

Solution

1. Step 1: Understand blitting update steps

   After drawing the updated artist, you must call fig.canvas.blit(ax.bbox) to update the screen.

2. Step 2: Analyze options

   You forgot to call fig.canvas.blit(ax.bbox) after draw_artist correctly identifies the missing blit call. Options A and C describe incorrect method orders. You did not call plt.show() at the end is unrelated if running in interactive mode.

3. Final Answer:

   You forgot to call fig.canvas.blit(ax.bbox) after draw_artist -> Option D

4. Quick Check:

   Missing canvas.blit call stops visual update [OK]

Hint: Always call canvas.blit after draw_artist to update [OK]

Common Mistakes:

• Not calling canvas.blit after draw_artist
• Calling copy_from_bbox too late
• Confusing restore_region order
• Assuming plt.show fixes blitting updates

5. You want to animate a scatter plot with 1000 points updating their positions in real-time. Which approach using blitting will give the best performance?

hard

A. Only update the figure title text each frame using blitting

B. Redraw the entire scatter plot from scratch each frame without blitting

C. Save background with copy_from_bbox, update scatter offsets, restore background, draw scatter, then call canvas.blit

D. Use plt.pause() inside a loop without blitting

Solution

1. Step 1: Identify efficient blitting steps for animation

   Best practice is to save the background once, then restore it each frame, update only the scatter points, draw them, and call canvas.blit.

2. Step 2: Compare other options

   Redraw the entire scatter plot from scratch each frame without blitting redraws everything, which is slow. Only update the figure title text each frame using blitting updates only title text, not points. Use plt.pause() inside a loop without blitting uses pause without blitting, which is less efficient.

3. Final Answer:

   Save background with copy_from_bbox, update scatter offsets, restore background, draw scatter, then call canvas.blit -> Option C

4. Quick Check:

   Blitting updates only changed scatter points fast [OK]

Hint: Save background once, restore, update points, then blit [OK]

Common Mistakes:

• Redrawing entire plot each frame
• Updating only title text instead of points
• Using plt.pause without blitting for speed