画質 高画質

Image mangled using VQGAN+CLIP.
Prompt: "fluffy rainbow bunnies".
Starting image (by ) is quoted in this tweet.
i={50;150;500;1000}
Generation video added as a comment in this thread.
(I personally dig the i=50 result)
https://t.co/gzLCa3ghe3

6 44

Hi, use the text : "Vibrant dog torus" generates different, similar images
your image is spectacular

2 78

VQGAN + CLIP "Adult Simba by Ross Tran" starting image on left

2 57

Negative prompts are perfectly normal

1 6

👾💬 ['a fantasy world contained into a single glass sphere and surrounded by giants living in clouds']🤔

1 2

So I've been playing with that ai generated image thing (vqgan+clip)
It's a little too much fun.

What do you think I put in for the prompts?

0 1

Larger resolution 768x768 MSE VQGAN CLIP Text-to-Image
"a witch in the style of Da Vinci"
"an ink drawing of chewbacca"
"pixel art of new york city in the style of Claude Monet"
"the australian outback"
https://t.co/3nUcgIS6of

2 20

Larger resolution 768x768 MSE VQGAN CLIP Text-to-Image
"a happy family in the style of Bob Ross"
"a marble sculpture of darth vader"
"a pastel of frogs"
"a vampire in the style of Frida Kahlo"
https://t.co/3nUcgIS6of

1 14

Larger resolution 768x768 MSE VQGAN CLIP Text-to-Image
"a babbling brook"
"a bronze sculpture of a fantasy land"
"a desert oasis in the style of Henri Matisse"
"a frog in the style of Hieronymus Bosch"
https://t.co/3nUcgIS6of

0 11

Never ignore a gut feeling, but never believe that it's enough. - Robert Heller | Flickr, storybook illustration
https://t.co/V1sXvhe3Zw

1 2