Ginga is written entirely in Python, and only uses supporting Python packages. There is nothing to compile (unless you need to compile one of the supporting packages).

In recent Linux, Mac, and Windows versions, all of the packages are available in binary (installable) form. It should not be necessary to compile anything, but as always, your mileage may vary.


  • python (v. 3.9 or higher)

  • setuptools-scm

  • numpy (v. 1.14 or higher)

  • astropy

  • pillow

Strongly recommended, because some features will not be available without it:

  • scipy

  • exifread

Optionally, for opening FITS files Ginga can use:

  • astropy (requirement)

  • fitsio

For WCS resolution you will need one of the following packages:

  • astropy (requirement)

  • kapteyn

  • astLib

  • starlink

BACKENDS (one or more)

Ginga can draw its output to a number of different back ends. Depending on which GUI toolkit you prefer (and what you want to do), you will need at least one of the following:

  • QtPy (PyQt5 or PyQt6)

  • PySide (pyside2 or pyside6)

  • pygobject (gi) AND pycairo (GTK 3)

  • tkinter

  • matplotlib

  • tornado

  • aggdraw

  • Pillow (PIL fork)

Notes on Supported Widget Sets

In the discussion below, we differentiate between the Ginga viewing widget, such as used in the examples/\*/example\*.py programs and the full reference viewer, which includes many plugins (ginga).


For the full reference viewer, Mac and Windows users should probably install the Qt version, unless you are the tinkering sort. Linux can use either Qt or GTK fine.


Ginga can use either PyQt or PySide, for Qt version 5 or 6. It will auto-detect which one is installed, using the qtpy compatibility package. There is support for both the basic widget and the full reference viewer.


If you have both installed and you want to use a specific one then set the environment variable QT_API to either “pyqt” or “pyside”. This is the same procedure as for Matplotlib.


Ginga can use GTK 3 (with gi). (If you have an older version of pycairo package, you may need to install a newer version from pycairo).


Ginga’s Tk support is limited to the viewing widget itself. For overplotting (graphics) support, you will also need one of:

  • Pillow

  • opencv-python

  • aggdraw


Ginga can render directly into a Matplotlib figure. Support is limited to the viewing widget itself. Any of the backends that Matplotlib supports is usable. Performance is not as good as to one of the “native” backends listed above, but oh, the overplot options!

HTML5 web browser

Ginga can render into an HTML5 canvas via a web server. Support is limited to the viewing widget itself. See the notes in examples/pg/ Tested browsers include Chromium (Chrome), Firefox, and Safari.

Basic Installation

You can download and install via pip by choosing the command that best suits your needs (full selection is defined in setup configuration file ):

pip install ginga  # The most basic installation

pip install ginga[recommended,qt5]  # Qt5

pip install ginga[recommended,gtk3]  # GTK 3

Or via conda:

conda install ginga -c conda-forge

The reference viewer can then be run using the command ginga.

Installation from Source

  1. Clone from Github:

    git clone

    Or see links on this page to get a ZIP file or tarball.

  2. Unpack, go into the top level directory, and run:

    pip install .

Platform Specific Instructions

Linux (Debian/Ubuntu)

If you are on a relatively recent version of Debian or Ubuntu, something like the following will work:

apt install python3-ginga

If you are using another distribution of Linux, we recommend to install via Anaconda or Miniconda as described below.

Mac/Windows/Linux (others)


For Mac/Windows or other Linux users, we recommend installing the Miniconda distribution (or Anaconda).

After installing Miniconda, open the Anaconda Prompt and follow instructions under Basic Installation via conda.

Running tests

  1. Install the following packages:

    $ pip install -e .[test]
  2. Run the tests using pytest:

    $ pytest

Building documentation

  1. Install the following packages:

    $ pip install -e .[docs]
  2. Build the documentation using make:

    $ cd doc
    $ make html