How to run Flux1 Locally In SD Forge And Integrate A 1111 Models plus Discover Tons Of Free AI Apps

preview_player
Показать описание
How to Run Flux.1 using SD and how to Integrate automatic 1111 models with SD Forge. Run Free Flux.1 Online.

Argument : --xformers --medvram --forge-ref-a1111-home
Рекомендации по теме
Комментарии
Автор

I have a 3060 12gb gpu like you.., and a 4070ti super 16gb gpu...when running flux 1_dev fp8 model my 4070ti super uses 13.2gb..so i am guessing my 3060 will struggle alot..or crash..dev 1 bnb nf4 works very well with my 3060 and dev 1 fp8 is fine for my 4070ti super...it is always better..and faster if you use only 80 percent of vram compared to running 95 percent vram...makes sense...having more system ram..say 32gb instead of 16gb with help alot for off loading from vram to system ram...64gb is my next purchase to replace my 32gb system ram...DPM++2M is the best sampler for realistic images...EULER is great for semi-surrealistic images i think...great video..thanks

kdzvocalcovers
Автор

if you choose > Diffusion in low bits> Automatic (fp16LoRa), loras will work

captainpike