... well, they look that way but no girls and no mess was involved in 'shooting' these photos. They are entirely generated by AI.

AI generated images from text prompts are definitely improving. With some actual training of WAM images by the model I think we could have something that can produce some amazing stuff. If I can lay my hands on a recent nvidia workstation card then I might give it a go.

Interested to know what you think of the images, they are definitely not perfect yet but I think they are looking pretty good.

Some are amazing!

Did you use Dall-E 2 or something else?

But this means no one actually got slimed .

Something else. These were done using Stable Diffusion https://stability.ai/blog/stable-diffusion-public-release

If you've got a reasonably powerful GPU you can run it on your own computer https://github.com/AUTOMATIC1111/stable-diffusion-webui

Nice! I tried Midjourney and Wonder and got some freaky results.

Like, Tim Burton does Lovecraft freaky...

I legit felt lied to.

I liked them.

Awesome pics!

How did you get the faces and the slime to look so good?

I tried stable diffusion but kept getting awful uncanny faces and abstract shapes for slime and pies.

There is a simple check box for face fixing. These were first generated through txt2img and then run through img2img with 16 images produced for each, I picked the best of the 16 for each. Each had slightly different prompts. I'm not sure if the online versions can do all these things as I installed a local version running on my GPU.

I find slime can look ok, pies did not look good from the attempts I made.