toot.wales is one of the many independent Mastodon servers you can use to participate in the fediverse.
We are the Open Social network for Wales and the Welsh, at home and abroad! Y rhwydwaith cymdeithasol annibynnol i Gymru, wedi'i bweru gan Mastodon!

Administered by:

Server stats:

586
active users

#aiart

96 posts48 participants6 posts today

I use image generators powered by latent diffusion models like Stable Diffusion or Flux for making rather random #elektrodada images filled with machine hallucinations, and cute mass-produced kitsch images just to test how good the machines have become at making those. I don't treat those images as "art", even though I use the #aiart tag for visibility; I treat them as anti-art, and I think this rotten world deserves anti-art, not bringing more beauty into the world but distilling the ugliness and the damage in a semi-automatic process. And for the beauty that we all crave in this ugly world, I generate sweet mass-produced nothings, cute but meaningless, like cheap sticky candy for the soul.
If anything, the AI models themselves are pieces of art, not the output they create. With a sufficient amount of human work, some output images can be refined into art though, but I'm rarely interested in that. I just drop some cryptic or silly prompts into the machine and see what happens.

I wonder why so many bloggers put AI images into their blog posts even when they aren't writing about AI. First of all, most of those texts don't need and images at all, and I remember how back in the 1990s, most of the first blogs didn't use any images at all in order to save bandwidth. Are you trying to beat the competition and catch more eyeballs? In that case, automatically generated images probably aren't going to work in the long run, since the novelty is wearing off.
Then again, there are some images that actually illustrate what the text is talking about, but in the time you spend polishing your prompt and running it through the image generator until you're satisfied with the result, you might have been able to find a CC-licensed stock image or even a public domain one of much better quality.

When I really need an image for something, I open Gimp or Krita or Inkscape and start working on it. I might include parts of AI images generated earlier, I might also include parts of images found on the web or in a magazine or in a still image from a video or parts of my own photos, whatever. I don't believe in "intellectual property", and I don't care if anybody finds an image of mine, even one that I made 100% from scratch without using the tiniest bit of any found media, and takes it apart in order to make something new out of it, uses it as a seed image or guide image or image prompt for an image generator, or puts it in the training data for a new diffusion model.

I actually like #AI. What I fucking hate is Capitalism. I don't want this stuff put into everything because people think it will make or save them money. I don't want big companies to build huge proprietary AI models that need a supercomputer to run. What I want is small open source AI models that can run on consumer hardware at home, no Internet connection needed, models that aren't owned by anybody and can be used by everybody for anything. And I want bloody crapitalism to crash and burn ASAP. If you see any images in my feed that you like, whether handmade or generated, do with them whatever you like, no need to ask for permission (although I do enjoy when people give me credit for my human-made stuff like my #pixelart, pencil drawings, or acrylic paintings).