Experimental LoRA.
Really small amount of images to train the LoRA on, so might not be great
Main tag:
lolammAppearance tags:
under eye bags, missing teeth, big hair, brown hair, hair over one eyeNegative tags:
Colored images used were almost all "flat colored" so if you want more vibrant looking results add to neg prompt
flat colorModel mostly, trained with black and white images so might need to add:
greyscale, monochromeDescription
FAQ
Comments (11)
I love that farm over there.
Yeah, same here. Shame theres not that much of it. I think the only other character that might be viable to make a LoRA of is Elizabeth.
Works great, I would love to see more Farm over there characters
Would love that as well, but might be difficult. Theres not that much images to work with. Had to do quite a lot of editing on these images already
I have 9 images for both Micaela and Elizabeth. Which one would you prefer more?
@Qneez Tough choice, probably Elizabeth, what would you say is the minimum image count for it to give decent results? The Lola's gives very consistent gens
@mcmanus tbh, I have no idea. I used 12 images for Lola. Like you said, pretty decent so far
@Qneez I'd be interested in helping clean up the images for use as Loras, I do occasional photo restoration.
@Loviatar If you are willing to help, sure, great! Its nothing complicated, just clean up some pics or do better line art, for some images. Quite a lot of images have this sketch style that probably are fairly bad to train a LoRA with.
Made a new one :)
THIS IS AMAZING!!! I LOVE IT!!