Holy crap! Have you guys seen the next generation of AI base model, I just saw the images people been making on stable diffusion sub reddit. Gonna be interesting to see if we get a pony model or NSFW finetunes. I downloaded the base models last night, 22gb or something like that. What do you guys think? Have you tried it yet?
-
Staff Applications are OPEN! [ Staff / Moderator ] More Info HERE Help us make a better forum for everyone!
New Flux model, what do you guys think?
- Thread starter Cooffrree
- Start date
There is now smaller models of flux that I played with, its SFW, but starting to pop up some NSFW loras on civitai. I use a 3060 12gb vram, works alright, takes some time to load XDA 22 GB model? I wish.
I'll be getting a nice shiny, new 4090, but not until November. I'll check back with you then
I did grab the microsoft PHI models and plan to test out the small ones this weekend.
Thanks! I grabbed this workflow: https://openart.ai/workflows/onion/flux-12gb-vram/JvCPHVvusg82HfGkOx9E
The page gives links tot he models and says where to put them. I'm generating the onion sample, right now
These looks amazing in quality, what prompt did you use? and any refines with sdxl or anything like that?
I tried some, but the people comes out "plastic" looking, dont get how people have these super good quality renders from it
The prompt for the one above is :- (my character lora), blue eyes, lightened brown hair, perfect nipples, pores, vellus hair, natural skin,, in the opulent rooms of an Arabian sheikhs palace, the harem girl wears belly dancer outfit. She is covered in gems and jewels and opulent rugs lie scattered around the room. Flux1-dev model, 1344x768, steps 20, cfg 1.0,
I've included the png if you want to grab the details from it (and if the forum don't strip them from images!)
No refines or the like were used, nor Adetailer