Quote:
Originally Posted by Ser-Goblin
Very very impressive! Can you share what kind of prompt and models you have used?
|
The images are generated using DALLE 3, available (and exploitable) at bing.com/create. But I'm mostly just querying their API illegaly. You just need a bit of technical skill to write a script and prompting skill.
DALLE is superior from a compositional perspective to Diffusion. While you could probably achieve equal or better results with Diffusion, it would take 20-30 times longer per image.
I'll attach a couple more, since people seem interested and have DM'd me.