osaic 2.0.0

Mercurial

From sources:

cd /wherever/you/want
hg clone https://bitbucket.org/iamFIREcracker/osaic
python setup.py install

PyPI

From sources:

cd /path/to/workspace
wget http://pypi.python.org/packages/source/o/osaic/osaic-2.0.0.tar.gz
tar zxvf osaic-2.0.0.tar.gz
cd osaic-2.0.0
python setup.py install

From the PyPI:

pip install osaic

Standalone application

A typical usage of the application is to display a mosaic composition created from a source image:

python -mosaic image.jpg

If you want to save the output to a file instead of showing it:

python -mosaic image.jpg -o mosaic-image.jpg

Finally, if you want to create a mosaic which is 4 times bigger than the original image, and with 100 tiles per side, just issue:

python -mosaic -z4 -t100 image.jpg

For everything else use the help message:

python -mosaic -h

Library

The module is a collection of objects and functions with different capabilities: functions for vectors, color transformations, image objects, image indexes and procedure wrappers.

Regarding operation with vectors, the module implement some basic functions not included in standard Python but useful while working with colors:

>>> dotproduct([1, 2, 3], [4, 5, 6])
70
>>> difference([1, 2, 3], [1, 2, 3])
[0, 0, 0]
>>> squaredistance([1, 2, 3], [0, 0, 0])
30

It is possible to find also a couple of functions wrapping up common colors operations, like computation of the average color of an image and color quantization. The latter is particularly useful while trying to keep the CPU work load at low levels:

>>> average_color('almost-red.png')
(240, 10, 20)
>>> quantize_color((240, 10, 20), levels=2)
(192, 64, 64)

As noted earlier, the module is built on top of the Python PIL library. However, we chose not export such external objects, but rather present to users some wrappers, namely ImageWrapper:

>>> img = Image.open('foo.png')
>>> img.size
(640, 480)
>>> img.reratio(5 / 1)
>>> img.crop((0, 0, 10, 10))
>>> img.size
(10, 10)
>>> img.show()

While creating mosaics, it comes in handy to have to possibility to index a set of images and make it possible to search which of them is the most similar, in terms of average color, to another one. The ImageList object is shipped with the module for this reason:

>>> img_list = ImageList(['1.png', '2.png', '3.png'])
>>> img_list.search((255, 0, 0))
ImageTuple(filename='1.png', color=(255, 0, 0), image=None)

Finally, the module is shipped with a mosaicify function which wraps up all the operations needed to create mosaic, including source images indexing, and search of neighbour images depending on the average color:

>>> import osaic
>>> osaic.mosaificy(
...     target='foo.png',
...     sources=['bar.png', 'asd.png', 'bazinga.png'],
...     tiles=128,
...     zoom=4,
...     output='mosaic.png',
... )