Stable Diffusion, NovelAI, Machine Learning Art - AI art generation discussion and image dump

OIG (3).jpeg
I can't believe they're releasing Bloodborne on the Switch!
Ignore the buttons

Edit:
OIG.Kw1R2yFKR7l.jpeg
What kind of monstrosity did I create? It was supposed to be a kiwi protecting infants from a pitbull.
 
Last edited:
What's with the memes with dogs and parachutes? I don't get it.

View attachment 5405265
Inspired by Hamas using paragliders, paramotors, etc. to break into Israel. Also the recent uproar over UK trying to ban the American Bully XL pitbull breed. It's the perfect representation of death from above.
 
I've heard recently that simply misspelling celebrities names is enough to potentially get by the filter yet still generate said person, as long as it's not too far off the mark. Haven't tried that yet but it would make sense.
 
I personally use SD.Next for sd1.5 stuff and comfyUI for SDXL. That is mostly a case of SDXL running like shit on automatic based stuff when i tried like 3 months ago. InvokeAI is looking more interesting so i may give that a go.

tldr
It all comes down to personal preference really. If a1111 works for you keep using it. I jumped to SD.Next because automatic was taking a break earlier this year and there were like 2 months with no updates. If you can get your head around comfyUI then you can do crazy things. A lot of the video stuff seems to be happening there. Though you need a lot of hardware for video.
View attachment 5402790
Thanks for the InvokeAI suggestion, it’s by far the best of these UI’s for local models. I tried using a few others but this one has great UI/UX.
 
Back