During the 2014 nanogenmo, I was caught out by a lack of tools. It's an excuse, and not a very good one. Other people managed to produce interesting work without writing much code. But not me.
In any case, as I looked through the completed nanogenmo projects, I was taken by projects which used images in their created texts, and especially by Generated Detective: A NaNoGenMo Comic, which featured a series of images which had been transformed to look like comics.
This seemed too cool for words, so I set about trying to do similar transformations. I dorked around with several different versions before rolling everything into one script driven by one massive config. Some samples (all originals courtesy of Wikimedia):
Churchill, the young warrior:
An unusual view of the Jefferson Memorial:
I'm still not happy with the process. I'd like to simplify the software requirements. If I had passed the Churchill through some sort of histogram equalization, I might have been able to render some of the details that are hidden in darkness. And I'd like better control over the colors, something that seems difficult without some sort of a gui.
The code is available here.