Nota : Pony being based on SDXL, my pony / SDXL models can work on SDXL / Pony checkpoints... or not, and in sometimes surprising ways. You have to try. Sometimes, a Pony/SDXL lora can even work on SD 1.5, but of course isn't made for good results like that.Here's my version of Lilith Calah. Cleo is a character from a comic called Dreamkeepers.Trigger : The trigger is lilithcalah.As the choice of training images is poor, you will have better results by adding more prompts used by the training.<hr />Change the word COLOR for the color you want, you silly goose ! Or just don't give color at all. Looks at my examples ! Honestly, look at the examples. Do you know the time it took me to do them on all those checkpoints ? I don't have a 600$ graphic card !the word color is meant to be replaced by the color you want. Throw as much paint on the poor girl as you want to !<hr />Prompting :The points here are captioned (used in the training, meaning have strong effect when you use them in your prompt) :COLOR hair : The color of the hair. By default, Lilith has green.COLOR body fur : She's supposed to be green, but often insist to give her an orange body fur if you don't prompt green body fur (or any other color - but any other of her prompt tags can often make her green even when it's not the body fur, for... reason. Mysteries of LoRa training...).COLOR eyes : By default, they are generaly purple, but can be red.cat tail is captioned. Sometimes you will need to add this, but can add another tail for... reasons.Other prompt :COLOR belly is captioned. But rarely works.All images have also been captioned with anthro and furry. Will help on Anime checkpoints, maybe also on furries, you try. Particularly true on Pony.Her facial mark on her face can appears on not, depending on the checkpoint. SDXL seem the be the better at it.Note for Pony :Remember : With Pony, it's better to start your prompt with :Positive (normal) prompt:score_9, score_8_up, score_7_up, score_6_up, source_furry BREAKThe BREAK can be ignored or not. Depend of your needs and your checkpoint, honestly.Negative prompt :score_4, score_3, score_2, score_1That can also help.Note for SDXL :anthro, furry in the prompt can be necessary on some checkpoints, or can make it TOO MUCH furry. You have to try.Tips:Basically, the more detail you add in your prompt, the better Lilith will be.This is because the few number of images I could find for the training dataset.clip skip can be 1 or 2. You should try.Lora weight : between 0.7 and 1. depends of the checkpoint.Trigger : The trigger is lilithcalah. You can put no weight on it, or give it 1.1 or 1.2, particularly if you put a lower weight to the lora.Also :Generation with DPM++ 2M / SDE and Automatic / Karras works well20 - 25 sampling steps should be largely enough.Upscaling upgrade the image a lot. I use R-ESRGAN 4x+ Anime6B, hires steps 0, Denoising strength 0.35, upscale by 2, and no refiner.Older versions :Anylora (SD 1.5)

Lilith Calah (Dreamkeepers) SD 1.5 + Pony + SDXL

Retrained with better settings, and around 10 more images

Files with SHA256: 2ee90688a728b5d47bedda0e011294e0efe752b73627df517a1404e1882634d2