AI - New Stable Diffusion model got delayed and now this makes the rounds...
(media.kotakuinaction2.win)
You're viewing a single comment thread. View all comments, or full comment thread.
Comments (17)
sorted by:
As someone who is peripherally familiar with what stable diffusion is, but hasn't kept up to date on the news surrounding it, could you or someone else chime in regarding what type of censorship was present in 1.5?
The model doesn’t understand weapons like guns very well so you could surmise that the dataset didn’t include very many gun related images.
Violence, gore, things you would expect to show up in a dataset of online images like Pepe and Wojak, sex, and nudity in general are all lacking in it.