Python libimagequant Integrations

This library provides helper functions for converting between objects in the unofficial Python libimagequant bindings library and "image" objects in various popular Python libraries.

If you'd like to add support for integration with more libraries, pull requests are welcome!

Install with pip:

python3 -m pip install libimagequant_integrations

API Documentation

This package contains the following submodules, for converting to/from the indicated classes:

• libimagequant_integrations.PIL (for Pillow's PIL.Image.Image*)
• libimagequant_integrations.PySide2 (for PySide2.QtGui.QImage)
• libimagequant_integrations.PySide6 (for PySide6.QtGui.QImage)
• libimagequant_integrations.PyQt5 (for PyQt5.QtGui.QImage)
• libimagequant_integrations.PyQt6 (for PyQt6.QtGui.QImage)
• libimagequant_integrations.skimage (for scikit-image numpy arrays^†)
• libimagequant_integrations.png (for PyPNG png.Reader and png.Writer^‡)

I suggest you import using (for example) import libimagequant_integrations.PyQt5 as liq_PyQt5, to avoid typing an excessive amount. :)

Each submodule contains two functions, used to convert to and from the libimagequant.Image class:

to_liq(image: OtherImageClass, attr: libimagequant.Attr) -> libimagequant.Image

from_liq(result: libimagequant.Result, image: libimagequant.Image) -> OtherImageClass

(Replace OtherImageClass with the appropriate other image class.)

You can check libimagequant_integration's version number using libimagequant_integrations.VERSION and libimagequant_integrations.VERSION_STR, which are in the same format as the version number constants in the libimagequant bindings library itself. The version numbers follow semantic versioning. They begin with 1.0.0, and do not track libimagequant's version numbers at all.

* While Pillow claims to support images with RGBA palettes, this feature seems extremely buggy, almost to the point of being completely unusable. Therefore, the from_liq() function discards the alpha channel and returns an image with a RGB palette. If you do need the alpha channel information, you can obtain it manually from the libimagequant.Image object.

^† Give to_liq() an RGBA image (numpy array of shape (y, x, 4), with uint8-type elements) as its "image" argument. from_liq() returns two numpy arrays: one with pixel data (shape (y, x, 1), uint8-type elements), and one with the RGBA color palette (shape (n, 4), uint8-type elements).

^‡ Give to_liq() a png.Reader as its "image" argument. from_liq() returns a png.Writer and a bytes object; you can save the image using the Writer's write_array() method with the bytes data provided.

License

Licensed under the GNU GPL v3. See the LICENSE file for the full license text.