Due for peer code review: Tuesday, February 22, 8pm;
and for in-class presentation: Wednesday, February 23
Start with a concept keyword or phrase — for example: tree, food, work, peace, blue, New York City. Gather some images of this concept. You could use Google Image search, a site like Pinterest or Flickr, your own original images, or something else. You should have at least 5 images, but you could have many more. Because you are automating the process of generation here, working with many more images will not make the below steps any more difficult to follow. In fact I would highly encourage you to work with many images to fully illustrate the power of this exercise. Create a new folder with your keyword as its name and put all the images into it.
Create a program that takes one command line
argument. Assume this argument is the name of a
directory. Pass the argument in
to listdir() to get the list of
files in that directory, and
use random.choice() to randomly select 4 to 5
images from that — let's call these your "working
images."
Try to use some image filtering techniques to modify the images in some way. For example, you might set bright or dark pixels transparent. Consider experimenting with some of the pattern generation techniques from last week, but instead of setting pixel colors into a new, empty image, modify one or more of your working images. Remember that if you may be starting with an image like a JPG that does not have any alpha, you can first convert it into an image object that does:
new_img = original_img.convert("RGBA")
Then, create a new empty image, and collage your working
images into it. (You can
use paste()
or alpha_composite()).
Save this new image as a file.
Once you have this functioning, run it at least three times to create different visual representations of this concept.
Now create a second directory full of images for a new, second concept. Without modifying your program, run your program again on this directory simply by specifying the directory name as a command line argument. Run your code a few times again for this new concept to see representations of this concept.
Create a new folder in your student folder called "Unit 1,
Tutorial 4". Inside that, put your code file, your image
directories, and your output images. Clearly name your output
images so they correlate with the directory of images that
they're drawn from. (For example, if your concept and directory
were called new-york-city/, your files
might
be new-york-city-1.png, new-york-city-2.png,
etc.
If you are feeling a little stuck and need some help getting started, I created a Github Gist with a code snippet and some explanation to help you get started. Have a look at that and let Viyan or I know if you're still confused.
Have all of your material uploaded to your Google folder by Tuesday, February 22, 8pm.
I have assigned each of you a classmate whose work you need to have a look at before class on Wednesday morning. The idea of code review is an important one in software development. Many organizations that develop software require any new code to go through a process of peer code review before that code is published live. Have a quick look at this article that offers a guide to doing code reviews. What we are doing here is not precisely a code review, but it's inspired by that idea. Find your code review assignment here.
Since we only have 100 minutes for 18 students, we are going to
do very fast "lightning presentations". You will show us your
image directories and your output images, then open up your code
file. It will be the job of your peer reviewer to offer a quick
comment about your work. The comment can be about concept,
formal qualities, aesthetics, or other choices, but I would like
the reviewer to say at least one specific thing about
one part of the code while it is up on the screen. For example,
"on line 25, this really interesting for loop is
responsible for the square pattern." This is an exercise in
reading code, but also talking about it — feel free to
contact your reviewee via email or with a Github Gist if you
have a question about their work.