TITLE: A Cool Project For Manipulating Images, Modulo Wordle
AUTHOR: Eugene Wallingford
DATE: March 16, 2022  2:52 PM
DESC: 
-----
BODY:
 Wordle has been the popular game <em>du jour</em> for a while now.
 Whenever this sort of thing happens, CS profs think, "How can I
 turn this into an assignment?"  I've seen a lot of discussion of
 ways to create a Wordle knock-off assignment for CS 1.
 
 None of this conversation has interested me much.  First, I rarely
 teach CS 1 these days, and a game such as Wordle doesn't fit into
 the courses I do teach right now.  Second, there's a certain
 element of Yogi Berra wisdom driving me away from Wordle:  "It's
 so popular; no one goes there any more."

 Most important, I had my students implementing Mastermind in my OO
 course back in the 2000-2005 era.  I've already had most of the fun
 I can have with this game as an assignment.  And it was a popular
 one with students, though challenging.  The model of the game
 itself presents challenges for representation and algorithm, and
 the UI has the sort of graphic panache that tends to engage students.
 I remember receiving email from a student who had transferred to
 another university, asking me if I would still help her debug and
 improve her code for the assignment; she wanted to show it to her
 family.  (Of course I helped!)

 However I did see something recently that made me think of a cool
 assignment:
<a href="https://twitter.com/5x6art">
   5x6 Art</a>,
 a Twitter account that converts paintings and other images into
 minimalist 5 block-by-6 block abstract grids.  The connection to
 Wordle is in the grid, but the color palette is much richer.
 Like any good CS prof, I immediately asked myself, "How can I
 turn <b>this</b> into an assignment?"

<center>
<a href="https://twitter.com/5x6art/status/1503350184629129224">
   <img src="http://www.cs.uni.edu/~wallingf/blog-images/art/wordle-art.png"
        alt="a screen capture of 5x6art from Twitter"
        width="400"
        align="center" />
</a>
</center>

 I taught our intro course using the media computation approach
 popularized by Mark Guzdial
<a href="http://www.cs.uni.edu/~wallingf/blog/archives/monthly/2006-08.html#e2006-08-31T17_12_04.htm">
   back in 2006</a>.
 In that course, my students processed images such as the artwork
 displayed above.  They would have enjoyed this challenge!  There
 are so many cool ways to think about creating a 5x6 abstraction
 of an input image.  We could define a fixed palette of <em>n</em>
 colors, then map the corresponding region of the image onto a
 single block.  But how to choose the color?

 We could compute the average pixel value of the range and then
 choose the color in the palette closest to that value.  Or we
 could create neighborhoods of different sizes around all of the
 palette colors so that we favor some colors for the grid over
 others.  What if we simply compute the average pixel for the
 region and use that as the grid color?  That would give us a
 much larger but much less distinct set of possible colors.  I
 suspect that this would produce less striking outputs, but I'd
 really like to try the experiment and see the grids it produces.

 What if we allowed ourselves a bigger grid, for more granularity
 in our output images?  There are probably many other dimensions
 we could play with.  The more artistically inclined among you
 can surely think of interesting twists I haven't found yet.

 That's some media computation goodness.  I may assign myself to
 teach intro again sometime soon just so that I can develop and
 use this assignment.  Or stop doing other work for a day or two
 and try it out on my own right now.
-----