This LoRA is for Inpainting only, it will not assist in generation of new images! Please read the directions!
BP, why the hell did you even do this? are you stupid?
So I was prodding with prompts to see what stupid things I could slap together when I noticed that even at denoising strength as low as .05, as well as using clashing colors, Pony will try to make cosmetics remain stable and will slowly warp colors till they are of a single tone.
Therefore the obvious thing to do when met with this meaningless problem was to force it to be more open minded, and here we are!
This LoRA will up that denoising strength .05 problem to as high as .4, though the actual size of the lips/eyeshadow as well as the pair of colors you choose to use will dictate if you should go as low as .1, and develop the area over several small generations.
Directions and Notes
Simply take your image, preferably lips or eyeshadow included, recolor the top and bottom halves in sketch by doodling over the original, maybe a few white dots or smears of a lighter shade for gloss, plop it in Inpaint, add "two tone lipstick/eyeshadow, (color) upper lip/eyeshadow, (color) lower lip/eyeshadow", and pray. Like colors will very likely mesh together; no red plus dark red, you gotta do it gaudy.
Your sampling method can play a big part in this...thing! And the .1-.4 denoising strength!
Depending on the size on lips, your sampling method may vary; thinner ones dont seem very fond of using Euler A, its sampling method causing the colors to blend at contact points, but it seems more consistent for this unorthodox request when using DPM++ 2M or DPM++ 3M SDE. In this aspect It requires multiple generations, maybe three, to find the sweet spot for whatever you are doing. While they do not take long due to the denoising being low, I will make sure you know of this prior to trying to generate large quantities of images all at once ultimately wasting time or money.
I only vaguely know why it can do the eyeshadow thing. Do not put too much faith in the eyeshadow thing.
This whole eyeshadow thing is purely happenstance. It is only very loosely tested, but helps with the whole two tone shtick as well as generating each eye with a different color eyeshadow while retaining consistency between their forms.
though much like the original goal I dont see much purpose yet barring being kinda funny.
Description
This LoRA was tested with PonyV6 and AutismMix
FAQ
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.