extcolors 0.2.1

1.1 Package

$ pip install extcolors

1.2 Manually

1. Download repository as zip.

2. Unpack zip into folder.

3. Enter folder.

4. Run the following command:

   $ pip install .

2.1 Input - Console

To use the application provide a path to the image that the application should extract colors from. In the following example the image is in the folder we are executing the command and the name of the image is gameboy.png:

$ extcolors gameboy.png

In this example gameboy.png refers to the following image created by Rebecca Machamer.

2.2 Input - Script

To use the application provide a path to the image that the colors should be extracted from. In the following example the image is in the folder we are executing the command and the name of the image is gameboy.png:

>>> import extcolors
>>> colors, pixel_count = extcolors.extract_from_path("gameboy.png")
>>> print(colors)
[((0, 172, 170), 386062), ((245, 245, 245), 59559), ((82, 90, 92), 17824), ((102, 184, 52), 15080), ((236, 27, 111), 1302), ((255, 180, 0), 137), ((241, 148, 185), 36)]

There is also the option to use an image already loaded through pillow.

>>> import extcolors
>>> import PIL
>>> img = PIL.Image.open("gameboy.png")
>>> colors, pixel_count = extcolors.extract_from_image(img)
>>> print(colors)
[((0, 172, 170), 386062), ((245, 245, 245), 59559), ((82, 90, 92), 17824), ((102, 184, 52), 15080), ((236, 27, 111), 1302), ((255, 180, 0), 137), ((241, 148, 185), 36)]

2.3 Output - Text

When the application is done it will output information about the execution. Probably the most relevant information is the [RESULT]-section that contains the extracted colors in RGB values and their occurrence rate presented in percentages.

Output based on gameboy.png:

Extracted colors:
(0, 172, 170)  :  80.43% (386062)
(245, 245, 245):  12.41% (59559)
(82, 90, 92)   :   3.71% (17824)
(102, 184, 52) :   3.14% (15080)
(236, 27, 111) :   0.27% (1302)
(255, 180, 0)  :   0.03% (137)
(241, 148, 185):   0.01% (36)

Pixels in output: 480000 of 480000

2.4 Output - Image

When the application is done it will also create an image in the directory where the command was executed. The image will contain the colors that where extracted sorted based on their occurrence rate, wrapping from from left to right. The image will use the name of the original image appended with a time stamp of when the execution took place.

Output based on gameboy.png:

4.1 Transparency

The support for images with transparency is limited. Colors that are fully transparent will be filtered out and will not be counted towards the colors in the result. Colors that have any level of transparency other than zero will be kept but the transparency will not be considered when comparing colors. If a more accurate result is desired the recommendation would be to apply a background color and perform a blend in an external application before extracting the colors.

4.1.1 Example - Full Transparency

The following image is 64 by 64 pixels large. The image consists of a border that is eight pixels wide and a fully transparent center.

Extracting colors from the image results in following where one can observe how the fully transparent pixels are removed from the percentage count.

Extracted colors:
(34, 32, 52)   : 100.00% (1792)

Pixels in output: 1792 of 4096

4.1.2 Example - Partial Transparency

The following image is 64 by 64 pixels large. The image consists of a border that is eight pixels wide and a center that has the same color as the border but with the alpha value set to 50% transparency.

Extracting colors from the image results in following where one can observe how the semi transparent color has been combined with the fully opaque color as the transparency was disregarded when the two colors were compared.

Extracted colors:
(34, 32, 52)   : 100.00% (4096)

Pixels in output: 4096 of 4096

4.2 Performance

When an image contains a large amount of different colors, which most do, the performance slows to a halt. If the grouping of colors is not desired/required then a workaround is to set the tolerance levels to zero. Setting the tolerance to specifically zero will make the application skip any comparisons from being made and become a simple counter resulting in much greater speeds.

Example, an image (3840x2160) containing about 340k unique colors will take two hours to complete with a tolerance level of 32 (the default value). However with a tolerance level of zero it will take ten seconds.