Week 3 - Edgelord Detection and review of creating Convex Hulls
0We went through a student's convex hull code and fixed the problems with it. Remember, we're using a really basic and inefficient algorithm here. In reality you should use Graham's or the Gift Wrapping algorithm, which are much, much faster and reliable. But the naive implementation is easier for students to write, so.
We then talked about edge detection, which is useful in many different types of situations, including compression (I showed how the Wavelet transform worked vs. JPEG), highlighting edges in video games to make characters pop out more (Borderlands does this well), cell shading (Okami would turn edges into strokes of ink and the area inside the edges into a flat color or two), computer vision, feature recognition, and so forth. The simplest edge detection algorithm very easy to write, just compute the derivative (change) from one point to the next. But they get a lot more interesting and complicated from there. (Again, I'm only expecting students to write very simple code here.)
Your next homework assignment is to turn the Voroni region generator into a clickable map within the terminal, and to only draw the outlines of the regions, rather than coloring in the whole region. If this is done automatically (without user input clicking on things) it can generate much more interesting countries than normal Voroni maps, which can be kind of boring.