Matplotlib中rcParams忽略特定默认FigureSize的问题求助

Let's break down why your (15, 9.3) default figure size is being ignored while extreme values work, and walk through actionable fixes.

Common Causes & Solutions #

1. Timing of rcParams Configuration #

Matplotlib loads its default settings when you first import pyplot. If you set rcParams after importing plt, initial setup might already have locked in a default size. Fix this by setting parameters before importing pyplot, or explicitly update them immediately after:

# Option 1: Set before importing pyplot
import matplotlib
matplotlib.rcParams['figure.figsize'] = (15, 9.3)
import matplotlib.pyplot as plt

# Option 2: Update right after importing (no plotting code runs first)
import matplotlib.pyplot as plt
plt.rcParams.update({'figure.figsize': (15, 9.3)})

Verify it works by checking the figure size right after creating a plot:

plt.plot([1,2,3])
print(plt.gcf().get_size_inches())  # Should return (15.0, 9.3)
plt.show()

2. GUI Backend Auto-Adjustment #

Interactive backends like TkAgg or QtAgg often adjust window sizes based on screen resolution or OS display scaling (e.g., Windows 125% scaling). This can override moderate figsize values but won’t constrain extreme sizes.

• Test with a non-interactive backend to rule out GUI interference:

  import matplotlib
  matplotlib.use('Agg')  # Saves to file instead of opening a window
  matplotlib.rcParams['figure.figsize'] = (15, 9.3)
  import matplotlib.pyplot as plt
  
  plt.plot([1,2,3])
  plt.savefig('test_figure.png')
  # Check dimensions: 15*100 = 1500px wide, 9.3*100 = 930px tall (default dpi=100)
  

• If the non-interactive test works, lock DPI to avoid scaling:

  plt.rcParams.update({
      'figure.figsize': (15, 9.3),
      'figure.dpi': 100,  # Fix DPI to match your desired physical size
      'figure.autolayout': False
  })
  

3. Third-Party Library Overrides #

Libraries like Seaborn or Pandas often modify Matplotlib’s defaults when you initialize their themes. If you use these tools, set your rcParams after their setup code:

import seaborn as sns
sns.set_theme()  # This may override figsize
plt.rcParams['figure.figsize'] = (15, 9.3)  # Re-apply your default

4. System Display Scaling #

OS display scaling (e.g., 125% on Windows, 1.2x on macOS) can make Matplotlib adjust figsize for readability. To bypass this, explicitly set the figure’s DPI to match your screen’s physical DPI:

import matplotlib.pyplot as plt

# Get screen DPI (example for TkAgg backend)
fig_manager = plt.get_current_fig_manager()
screen_dpi = fig_manager.window.winfo_screendpi()

plt.rcParams.update({
    'figure.figsize': (15, 9.3),
    'figure.dpi': screen_dpi
})

Final Troubleshooting Step #

If nothing else works, reset all Matplotlib settings to default before applying your figsize to eliminate conflicting configurations:

import matplotlib as mpl
mpl.rcParams.update(mpl.rcParamsDefault)  # Reset to factory defaults
mpl.rcParams['figure.figsize'] = (15, 9.3)
import matplotlib.pyplot as plt

内容的提问来源于stack exchange，提问作者oliversm