Given a picture and a number k, this script outputs a version of your picture that only uses k colors. But it picks those k colors wisely (using the k-Means Clustering Algorithm) so as to preserve the original context of the image. You'd be amazed how much trendier you look with fewer colors.

##How to run the script

1. Get the file entitled kMeansColor.py (can be done by downloading the zip file of this repo on github)

2. Install dependencies, of which there are two: SciPy and PIL (Python Imaging Library). If you're a Mac-user you can install these by simply entering the following into the Terminal (you'll be prompted for your computer password):

sudo port install py27-scipy
sudo port install py27-pil

3. Run the code by entering the following into your terminal:

python2.7 kMeansPainting.py [filepath to picture] [number k]

Note that your terminal needs to be in the folder containing kMeansPainting.py; if this is not the case, then enter cd [path to folder containing kMeansPainting.py] before running the above line of code.

When the script is complete, a new version of your image will appear containing only the most identifying k colors (it will NOT overwrite your original image). Most jpg/png files take ~2 seconds to complete, while high-resolution photos may take up to a minute (pending further testing).

4. Report errors that you encounter by letting me know (email: [email protected], twitter: @nikhilhyphen). I hacked this together quickly and haven't done much error handling or testing for corner cases, so let me know if you run into issues and I can update the code accordingly!

##Example ###Original vs. k=2: