I went on an extensive cats-with-guitars AI spree a couple of weeks ago. Although I’ve made other AI images since then, I apparently need to get a few of the cats out of my system before sharing the others.
I especially like the watercolor-y texture of the image below. Did you notice this cat is holding the guitar with its tail? It doesn’t make sense rationally but, at least to my eye, it does make sense visually.
The last image kinda sorta combines the first two. I personally think it looks a bit like Batman with whiskers.
On a related note, the Washington Post published an opinion piece about images created with Midjourney1 that the US Copyright Office ruled can’t be copyrighted. They say that “In cases where non-human authorship is claimed, appellate courts have found that copyright does not protect the alleged creations.”2
This is really frustrating. For one thing, it’s pretty clear to me that the word prompts I give Midjourney create results based on my own previous use of the app, with more recent uses having greater weight. Which means that anyone could use exactly the same word prompt and end up with a dramatically different result.
And furthermore, it’s quite difficult and time-consuming to get Midjourney to give me what I want or, failing that, simply something that I like. It almost always takes multiple iterations and tweaks of a prompt — hours of work — to get even roughly where I want to be. It’s hard for me to believe that my AI creations are any more “non-human” in authorship than are my photos.
All that said, the WaPo opinion piece makes a really good argument in favor of AI copyright.
Midjourney is my AI app of choice.
The Washington Post piece is here (the link bypasses their paywall legally); the Copyright Office’s decision is here. It’s much more readable than I would’ve believed possible, and does a very good job of describing how Midjourney works.
I love the kitty guitar players!!!