You are reading an old version of the documentation (v2.0.1). For the latest version see


Table Of Contents

This Page

What’s new in matplotlib

For a list of all of the issues and pull requests since the last revision, see the GitHub Stats.

New in matplotlib 2.0


matplotlib 2.0 supports Python 2.7, and 3.4+

Improved color conversion API and RGBA support

The colors gained a new color conversion API with full support for the alpha channel. The main public functions are is_color_like(), matplotlib.colors.to_rgba(), matplotlib.colors.to_rgba_array() and to_hex(). RGBA quadruplets are encoded in hex format as #rrggbbaa.

A side benefit is that the Qt options editor now allows setting the alpha channel of the artists as well.

New Configuration (rcParams)

New rcparams added

Parameter Description
date.autoformatter.year format string for ‘year’ scale dates
date.autoformatter.month format string for ‘month’ scale dates format string for ‘day’ scale dates
date.autoformatter.hour format string for ‘hour’ scale times
date.autoformatter.minute format string for ‘minute’ scale times
date.autoformatter.second format string for ‘second’ scale times
date.autoformatter.microsecond format string for ‘microsecond’ scale times
scatter.marker default marker for scatter plot
svg.hashsalt see note,, xtick.bottom, xtick.minor.bottom, xtick.major.bottom ytick.left, ytick.minor.left, ytick.major.left ytick.right, ytick.minor.right, ytick.major.right Control where major and minor ticks are drawn. The global values are and ed with the corresponding major/minor values.
hist.bins The default number of bins to use in hist. This can be an int, a list of floats, or 'auto' if numpy >= 1.11 is installed.
lines.scale_dashes Whether the line dash patterns should scale with linewidth.
axes.formatter.offset_threshold Minimum number of digits saved in tick labels that triggers using an offset.

Added svg.hashsalt key to rcParams

If svg.hashsalt is None (which it is by default), the svg backend uses uuid4 to generate the hash salt. If it is not None, it must be a string that is used as the hash salt instead of uuid4. This allows for deterministic SVG output.

Removed the svg.image_noscale rcParam

As a result of the extensive changes to image handling, the svg.image_noscale rcParam has been removed. The same functionality may be achieved by setting interpolation='none' on individual images or globally using the image.interpolation rcParam.

Qualitative colormaps

ColorBrewer’s “qualitative” colormaps (“Accent”, “Dark2”, “Paired”, “Pastel1”, “Pastel2”, “Set1”, “Set2”, “Set3”) were intended for discrete categorical data, with no implication of value, and therefore have been converted to ListedColormap instead of LinearSegmentedColormap, so the colors will no longer be interpolated and they can be used for choropleths, labeled image features, etc.

Axis offset label now responds to labelcolor

Axis offset labels are now colored the same as axis tick markers when labelcolor is altered.

Improved offset text choice

The default offset-text choice was changed to only use significant digits that are common to all ticks (e.g. 1231..1239 -> 1230, instead of 1231), except when they straddle a relatively large multiple of a power of ten, in which case that multiple is chosen (e.g. 1999..2001->2000).

Style parameter blacklist

In order to prevent unexpected consequences from using a style, style files are no longer able to set parameters that affect things unrelated to style. These parameters include:

'interactive', 'backend', 'backend.qt4', 'webagg.port',
'webagg.port_retries', 'webagg.open_in_browser', 'backend_fallback',
'toolbar', 'timezone', 'datapath', 'figure.max_open_warning',
'', 'tk.window_focus', 'docstring.hardcopy'

Change in default font

The default font used by matplotlib in text has been changed to DejaVu Sans and DejaVu Serif for the sans-serif and serif families, respectively. The DejaVu font family is based on the previous matplotlib default –Bitstream Vera– but includes a much wider range of characters.

The default mathtext font has been changed from Computer Modern to the DejaVu family to maintain consistency with regular text. Two new options for the mathtext.fontset configuration parameter have been added: dejavusans (default) and dejavuserif. Both of these options use DejaVu glyphs whenever possible and fall back to STIX symbols when a glyph is not found in DejaVu. To return to the previous behavior, set the rcParam mathtext.fontset to cm.

Faster text rendering

Rendering text in the Agg backend is now less fuzzy and about 20% faster to draw.

Improvements for the Qt figure options editor

Various usability improvements were implemented for the Qt figure options editor, among which:

  • Line style entries are now sorted without duplicates.
  • The colormap and normalization limits can now be set for images.
  • Line edits for floating values now display only as many digits as necessary to avoid precision loss. An important bug was also fixed regarding input validation using Qt5 and a locale where the decimal separator is ”,”.
  • The axes selector now uses shorter, more user-friendly names for axes, and does not crash if there are no axes.
  • Line and image entries using the default labels (“_lineX”, “_imageX”) are now sorted numerically even when there are more than 10 entries.

Improved image support

Prior to version 2.0, matplotlib resampled images by first applying the color map and then resizing the result. Since the resampling was performed on the colored image, this introduced colors in the output image that didn’t actually exist in the color map. Now, images are resampled first (and entirely in floating-point, if the input image is floating-point), and then the color map is applied.

In order to make this important change, the image handling code was almost entirely rewritten. As a side effect, image resampling uses less memory and fewer datatype conversions than before.

The experimental private feature where one could “skew” an image by setting the private member _image_skew_coordinate has been removed. Instead, images will obey the transform of the axes on which they are drawn.

Non-linear scales on image plots

imshow() now draws data at the requested points in data space after the application of non-linear scales.

The image on the left demonstrates the new, correct behavior. The old behavior can be recreated using pcolormesh() as demonstrated on the right.

(Source code, png, pdf)


This can be understood by analogy to plotting a histogram with linearly spaced bins with a logarithmic x-axis. Equal sized bins will be displayed as wider for small x and narrower for large x.

Support for HiDPI (Retina) displays in the NbAgg and WebAgg backends

The NbAgg and WebAgg backends will now use the full resolution of your high-pixel-density display.

Change in the default animation codec

The default animation codec has been changed from mpeg4 to h264, which is more efficient. It can be set via the animation.codec rcParam.

Deprecated support for mencoder in animation

The use of mencoder for writing video files with mpl is problematic; switching to ffmpeg is strongly advised. All support for mencoder will be removed in version 2.2.

Boxplot Zorder Keyword Argument

The zorder parameter now exists for boxplot(). This allows the zorder of a boxplot to be set in the plotting function call.

boxplot(np.arange(10), zorder=10)

Filled + and x markers

New fillable plus and x markers have been added. See the markers module and marker reference examples.

rcount and ccount for plot_surface()

As of v2.0, mplot3d’s plot_surface() now accepts rcount and ccount arguments for controlling the sampling of the input data for plotting. These arguments specify the maximum number of evenly spaced samples to take from the input data. These arguments are also the new default sampling method for the function, and is considered a style change.

The old rstride and cstride arguments, which specified the size of the evenly spaced samples, become the default when ‘classic’ mode is invoked, and are still available for use. There are no plans for deprecating these arguments.

Streamplot Zorder Keyword Argument Changes

The zorder parameter for streamplot() now has default value of None instead of 2. If None is given as zorder, streamplot() has a default zorder of matplotlib.lines.Line2D.zorder.

Extension to matplotlib.backend_bases.GraphicsContextBase

To support standardizing hatch behavior across the backends we ship the matplotlib.backend_bases.GraphicsContextBase.get_hatch_color method as added to matplotlib.backend_bases.GraphicsContextBase. This is only used during the render process in the backends we ship so will not break any third-party backends.

If you maintain a third-party backend which extends GraphicsContextBase this method is now available to you and should be used to color hatch patterns.