01-27-2024, 07:15 AM
Spoiler: (click to show)(click to hide)
it's because of the way AI is trained, when tagging images to teach concepts to AI, developers didn't tag it like "photo of Hillary Clinton, no elephants, no dogs, no cats, no potted plants, no cybertruck" ad nauseum, you only say what's there
so you can't say "none of this thing" because then you've included the word of that "thing" which is enough
in a power-user-friendly UI like with locally hosted stable diffusion, you can negative prompt, literally put elephants in a different text box and they will be actively excluded, but in simple interfaces like chatgpt and bing you can only say what you want, not what you don't want
so you can't say "none of this thing" because then you've included the word of that "thing" which is enough
in a power-user-friendly UI like with locally hosted stable diffusion, you can negative prompt, literally put elephants in a different text box and they will be actively excluded, but in simple interfaces like chatgpt and bing you can only say what you want, not what you don't want