A Merge of the High Quality models and some loras for fine-tuning.
Focused on high-quality 2.5d and semi real images.
FP8 VERSION USES WAY LESS MEMORY! Bringing this one to FP8 cut the size in half but did not lose any quality. Way better suited for not so high end GPUs than usual FP16 Checkpoints.
VAE required for v1.0, there is no VAE baked in! Otherwise images look like negatives
Description
FAQ
Comments (8)
Tried this with both Fooocus and InvokeAI but all results (although generating looks ok) the end result gets put out like a negative/inverted image
Ok, tried everything on Fooocus too and it's garbage. Ah well.
yes that is because noe VAE is baked into it. Gotta use one
@JohnnyB1 Damn, on Fooocus sadly can't use custom VAE...BUUUT, will try on InvokeAI to see if it works. Any VAE works?
@JohnnyB1 Changed the review :) ... Works fine if the VAE is added. Maybe add that in the description not to have people rub their heads not knowing why their output images come out bad ;)
@Lancerx Fooocus can use another VAE you just need to change it. I didn't realise until now. edit: tested thoroughly and it's NOT a vae issue. This and only this model so far doesn't like Fooocus. I've used plenty of models that also clearly say they work best with a vae and no issue whatsoever.
@rivdemon1221554 How do you change in Fooocus the VAE? I know there is the VAE folder but how do you make it to USE it? Also did you try disabling the 3 default styles? Fooocus V2 , Foocus Enhance and Fooocus Sharpen?
@rivdemon1221554 never used focus, I just know it works with a vae :/