Aren’t they just recycling old comics since there are so many of them? Or did some smart ML researcher create a language/ comic model based on a corpus of comics he wrote when he was alive, and some editors just generate and curate? I’m guessing we aren’t at the latter…yet.
I taught a senior capstone group who were generating new Calvin and Hobbes cartoons using AI in 2019. They weren’t great, but I’m sure today the results would be 1000% better.
What they did was run sentiment analysis on the corpus of comics, and then categorized all the panels by a number of emotions (anger, joy, surprise, confusion, etc.) then they wrote some jokes, specified some constraints about the panels, and then the algorithm assembled a comic.
Today I imagine they could generate entirely new panels from the corpus, and maybe ChatGPT could be trained on the text to write jokes closer to the style of Calvin and Hobbes.
This gives me a brilliant idea: maybe we could make an AI to make new movies based on old ones, and train it on the first 2 Alien movies, so we can finally get a decent sequel to that franchise. Humans post-~1990 have proven that they're completely incapable of the task.
