Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.
- 158
- 2
Jump in the discussion.
No email address required.
Notes -
AI-generated images are here - and they're awful.
Dall-E 3 seems to have crossed the line into "good-enough" territory that many smaller websites are now using AI-generated images. It seems, unfortunately, that Dall-E 3 has a style. And that style is awful. The scenes are always far too busy, everything is cartoony, and the colors are oversaturated.
It's the new Corporate Memphis.
It also makes me feel super racist that I have to constantly tell it to draw white people since it has an even looser grip on historical accuracy then Bridgerton.
Try running Stable Diffusion. It's open source and you can find a specifically-trained model for whatever you want. If you can't run it locally, you can find a million services online that can do it for you. You don't need to deal with content restrictions or it adding diversity to all of your prompts. I use a site called NovelAI, which runs its own version of SD trained off of an anime booru. They recently put out a new version of the image gen and it's really good. The progress made since just the last year is astounding.
We could have easily been in the timeline where the big three image diffusion models (Dall E, Stable Diffusion, and Midjourney) all kept it locked down, but since one of them broke ranks and made it open source, it's opened up a world of potential for everyone to use image generation how they want.
More options
Context Copy link
More options
Context Copy link