The Program

Recently I’ve started collecting images online (specifically off of the flickr website owned by Yahoo!).  My goal is to collect roughly 65,000 images to build a mosaic out of them.  I’ve only collected roughly 6000 so far.. so I don’t have nearly enough to began building a mosaic.

I have done one experiment on them, however.  I’ve averaged all the pictures together (Resize them to 256×256, summed each pixel up and divided by the number of images).  I obtained this:

As awesome as that is.. it appears mostly gray.  You can see some hints of blues and reds in there, which I found interesting.  I then took this, and simply normalized the image to obtain a more apparent version of how the colors interacted (Excuse the poor jpeg quality).

This is a much better image to describe what is going on.  My [limited] explanation for what is going on is that darker colors tend to be on the bottom (floor, ground, etc), while faces and people tend to be in the middle, and sky and clouds tend to be on top.

I intend on doing more experiments with these images once I collect more of them. And once I get enough, mosaic time!

(On a more technical note, the image fetchers and processors have been programmed in python, using MySQL as a backend to keep track of the histograms of every image.. to make access time for creating a mosaic much faster).

Today, I’m featuring gnuPlot.  gnuPlot is a windows and Linux application to make 2D and 3D graphs of different styles.  It’s a great application.  The only downside is the non graphical user interface (except for the actual display of the graph).  This means you have to learn how to use it’s feature rich language.  This isn’t a big deal as most of it is common sense (i.e: f(x)=2*x; plot f(x);).  Some of it, however, can be a bit more complex.  Of course, if you want to do more complex things, you have to learn how to use more complex parts of the program.

Overall, it’s a very good and well written program, and I suggest it to anyone looking to make advanced graphs.

One of my projects a month ago was to use gnuPlot to create a 3D representation of the perlin noise function.  A few hours later, and after learning how to use many of the more advanced mathematical routines in it’s built in function base, I accomplished this. Below, you can find the graph and the functions I used to display it.

Perlin Program

I got a little bored so I decided to have some more fun with inkscape.  I pulled up my trusty web browser and went over to xkcd.  Found a nice simple comic, and decided I wanted to blow it up.

The result?  A 1680×1050 (or any size for that matter) version of the traveling salesman problem.

Full credit to xkcd, of course.  Here it is! :