K@Dr. Furkan Gözükara Have you seen the Kalpana model? It's 32 GB instead of FLUX.DEV which is 23GB. It produces the same result with the same prompt... I have no idea what the deal is with it?! Do you know? https://huggingface.co/enhanceaiteam/kalpana
Fuses a lot of shared vram on big images but still works very fast
Fhow could they have trained in such low time weird
KThis is the division of the two images created with Kalpana and Dev. It seem to be some difference, but nothing visible in the images
Flol it is fake 
KIt works
K
KJust don't knwo what it is
FK
Fworks but same thing :d
KI wonder why it's 9GB larger
Fit has more billion paramets
Fnot making sense either haha
KWierd
KMore parameters should normallybe better right? Or could be?
Fit should be
Fbut i dont think they did any training
![Dazzastrous [4090]](https://cdn.discordapp.com/avatars/871389071686135838/c4a5f4d81d9eeb18d856295d692e2c07.webp?size=40)
Dfollow the prompt better , more detailed pics
Dthink the right one is slightly sharper or might be my eyes lol
KYour eyes!
Ddamn my eyes
KOr you have supersight. What do I know 
AAnyone test the loras for flux yet
Ddid you say theres been an update recently to InstantID doc?
![Dazzastrous [4090]](https://cdn.discordapp.com/avatars/871389071686135838/c4a5f4d81d9eeb18d856295d692e2c07.webp?size=16)
Fv19 is last
Fon gradio you will see
Fwe get the best realism with some config
Fbut clothing degrades
F
Dwhats the shadow behind ?
Fgradio bug :d
F
Fmaking runpod massed compute and kaggle to publishj on patreon
Fdmn A40 is cheap on runpod for some reason
F48 gb
APineapple cat lol
Adr have you trained loras for Flux yet
Fnot yet sadly
Fi am waiting kohya to finalize
Asimpletrainer I think some people are using
Ahave you tried it
Asimple tuner*
Fyes ,i follow it
Fbut i am waiting kohya
Fit is gonna use hugging face version
Flol look at vram
F
Flol filled entire shared vram haha
Fbut works great