Hello welcome!
Here I leave a small introduction of the LoRA:
Today I leave you a nice style with loose strokes ~ (￣ω￣)/
Femmes | IL V2.5
What the heck am I doing?!
Lately, I’ve been training LoRAs on Illustrious 2.0, and while they worked pretty well with other checkpoints, they sometimes didn’t work at all with mine.
At one point, I even started wondering if it was my fault—like, in my head, I thought that if I fused it with Illustrious 2.0, that would automatically make it compatible with LoRAs trained on it…
Turns out I was wrong. It actually works way better with LoRAs trained on IL1.0.
So… I guess that’s going to be my new base for LoRA training.
Sorry for the confusion 😅
Femmes | IL V2
I gave it another training pass using Illustrious 2.0, building on the results from the previous LoRA, as the original dataset was already used in version one. ~
Femmes | IL V1
I still remember this like it was yesterday, it all started with an order for a bountie, someone wanted a style that will make hearts on their cheeks and I didn't think twice, I did it.
Maybe I didn't win but what memories ~ XD
After trying it for a while I can say that I like it, at first I thought that the style had been lost and it would only be a drawn version without much definition (which is not bad but it was not at all similar to its first version).
But then I reviewed the dataset and chose the words that were most repeated, eliminated many prompts and that's it.
PS: Konan works so great with this, Illustrious practically became quite realistic.
Femmes | SD1.5
