Generative ai is fun, useful and sometimes crazy addictive!

 As you probably saw in the previous couple of posts already, I've been going hog-wild with generative AI's to generate cool-looking hindy-mythology images lately. And let me tell you what a big load of fun it is, you put in the words, give it the style, and modify the images until you get pretty much, or pretty close to what you want, within given constraints. I've generated about a couple of hundred images at this point, saved about a hundred, and ohhh man so addictive it's like playing a video game, I think I'll sit down for fifteen minutes and generate something cool, and I'm still there four hours later at three in the morning telling myself I'll generate just three more, three more.

It has also taught me the strict limits of the generative models, in that I can sort of see what they're doing, where the training data probably came from, what the labeling of the data probably looked like, and what sort of transformations were going underneath to make it all happen. It does feel a little bit disappointing once you know how the magician does it, but it's still going to be useful to create thumbnails for my upcoming vodcast, so that's all fine and dandy.

I imagine I'll be writing a lot more on the matter, so stay tuned!