A couple years ago I made a post with generative AI image-maker Crayion (aka DALL-E). Crayion has been upgraded since then, so let’s take another look at Crayion and see what it has learned about the k-pops in two years!
For this post I’m going to use exactly the same prompts as the original “Crayion Punk” post, so we can gauge improvement. Crayion has changed a little and now has a “style” prompt but I’ve switched that off because I don’t have any. Have fun comparing the two posts! For what it’s worth, I think that while Crayion has clearly become much better at drawing in 2024 compared to 2022, it’s actually simultaneously also somehow become much worse at understanding what I want it to draw. I think this might be an effect of the “AI collapse” problem scientists are discovering, where AI becomes increasing generic and unable to advance further, because it inevitably starts training on its own data and magnifying its own mistakes. So I don’t think the threat of AI taking over the k-pops is going to manifest soon.
Note that for those concerned about the environmental impacts of AI images, know that I held in my own farts for the entire duration of making this post, thus making sure methane (a potent greenhouse gas, much worse than carbon dioxide) didn’t add to the global warming effect. Stan Chuu.
BONUS BLACKPINK PAVED THE WAY CONTENT
EXTRA BONUS NEW CONTENT TO BRING THIS POST UP TO 69 IMAGES
That’s all for this post! Kpopalypse will return!
