Re:Google “We Have No Moat, and Neither Does OpenAI”
by jamilton on Thursday May 04, 2023 @05:47PM
(#35822557)
ControlNet helps a lot with composition and lighting (https://sandner.art/create-atmospheric-effects-in-stable-dif...). It requires more work than just entering a prompt, but probably less work than doing it manually once you get used to it. I think there's a number of StableDiffusion clients in development that are trying to make this easier.
[ Parent
| on HN ]
Re:Google “We Have No Moat, and Neither Does OpenAI”
by chefandy on Friday May 05, 2023 @02:45PM
(#35833541)
Ha... I accidentally replied to the wrong comment. Anyway, thanks! That's pretty neat. The sample images look a bit overwrought like a lot of other AI images do but I'll bet they're doing that to follow the trend rather than it being a technical limitation.