This makes Moralgear's OC Lilithia, a succubus who drains size to grow bigger. Suggested to be used at 0.6. This model was trained on illustrious-xl and generated in Illustrious pencil-xl. Made in local Kohya. Any suggestions to improve or recommended things for the data set would be appreciated. If you want to check out the artist you can find them <a target="_blank" rel="ugc" href="https://www.google.com/url?sa=t&amp;source=web&amp;rct=j&amp;opi=89978449&amp;url=https://x.com/moralgear%3Flang%3Den&amp;ved=2ahUKEwjsidbgs6mMAxVQIxAIHb44O5wQjjh6BAggEAE&amp;usg=AOvVaw1-cNfVl133v5YEZW6Tmorq">here.</a> Her default clothes are: crop top, detached collar, puritan collar, bridal gauntlets, miniskirt, hip vent, highleg panties, purple panties, tailcoat, purple shirt, purple thighhighs, thigh cutout stirrup legwear cat ears, demon tail Her human form is: wide hips, flat chest, denim shorts, tank top If anyone is aware if this character was drawn by anyone other than the original artist please tell me (I think she was in a few art trades, but I forget with who).

Lilithia (Moralgear OC) Illustrious

This was made using all the images I could find. Her collar I decided to call a 'detached collar, puritan collar'. Her tailcoat is separate from her wings since it shows still when when her wings are revealed. She is depicted pretty much the same amount with her tail as she is without it so the tag is separate. 2 versions of her have a logo on the crop top to make the dataset stable the 2 had the logo removed. Her human form only has 2 images both were line arts with no color so I have no idea what color her shirt is supposed to be. The cat ears are only there when her familiar is pretending to be her by shapeshifting. When she has no prey she eats vegetables (which is why I made an image of her eating carrots (the one image of this was a line art so I couldn't tell if they were radishes or carrots)). This was made using 3 tiers normal images x10, cropped head shots x8, outfit crops x6. Still working on the moralgear style retag, hand tagging that many images takes me awhile. I decided to prioritize one of the OC's to get a little lora going along the way. Probably will do the other OC's (that have enough pictures) when I finish the dataset.

Files with SHA256: 72b5af8b8fd6880daaba82779a699436d8f1bbd3b88076b224c94278150a2f99