[This is the nearest thing to Animagine XL you can get, since UrangDiffusion non-base version is finetuned further and broke some artist tags.]
Archive page for UrangDiffusion Base models (pretrained, non-finetuned versions).
For training details, please refer to UrangDiffusion main page.
The model falls under Fair AI Public License 1.0-SD license.
===
Description
Pretrained version of UrangDiffusion v1.1
FAQ
Comments (3)
I've always wondered, what's the actual use of pre-fine tuned versions like this one? Is there a specific scenario where I should prefer one or the other?
In UrangDiffusion case,
1. If you prefer the style closer to Animagine XL and don't mind accuracy.
2. If you wanna finetune the Unet yourself, or even continue the pretraining of the model.
But the reason I'm posting base is to archive the model, in case I need it later.
@kayfahaarukku Oh I see, thank you for the info! After playing a bit with it, I can't be 100% sure as I'm very amateurish, lol, but I feel like the base model responds much more strongly to artist names in the prompt, which I don't dislike at all! I posted a couple images with a very similar prompt, with just different styles!