tilequant 1.2.1

Installation

Python 3 is required.

Via pip3:

pip3 install -U tilequant

Usage

Usage: tilequant [OPTIONS] INPUT_IMAGE OUTPUT_IMAGE

  Converts any image into a indexed image containing a number of smaller
  sub-palettes (--num-palettes), each with a fixed color length (--colors-
  per-palette). The conversion will assign each tile in the image one of
  these sub-palettes to use (single-palette-per-tile constraint).

  INPUT_IMAGE is the path of the image to convert and OUTPUT_IMAGE is where
  to save the converted image. All image types supported by PIL (the Python
  image library) can be used. :return:

Options:
  -n, --num-palettes INTEGER      [Default: 16] Number of palettes in the
                                  output.
  -c, --colors-per-palette INTEGER
                                  [Default: 16] Number of colors per palette.
                                  If transparency is enabled, the first color
                                  in each palette is reserved for it.
  -t, --transparent-color TEXT    A single color value (hex style, eg. 12ab56)
                                  that should be treated as transparent, when
                                  doingthe conversion with transparency
                                  enabled.
  -v, --loglevel [DEBUG|INFO|WARNING|ERROR|FATAL|CRITICAL]
                                  [Default: INFO] Log level.
  --help                          Show this message and exit.

Examples

For the new version no examples exist yet. However to get a general idea, you can view the examples of the old version in “examples”.

Transparency

The actual amount of colors per palette is one lower than specified and a “transparency color” is added at index 0 of all palettes. If transparent-color is specified, the image is scanned for pixels with this color first and in the end, those pixels will be assigned their local “transparent color” index.

Legacy version

Originally (before integrating the new and much better newer version based on Tilequant by Aikku93) there was a pretty bad pure-Python based version of this tool. It is no longer available in current versions of this lib, to access it see versions prior to 1.x.

The only thing left over from the legacy implementation is Tilequant.simple_convert (only accessible via API) which allows trying to convert images without running any sort of quantization on them, failing if not possible.

Special Thanks

• Aikku93

• Nerketur

• AntyMew

• psy_commando