Character LORA of Kuroeda-san from the manga Elf-san Wa Yaserarenai by Synecdoche. Trained at 1024x1024 for 32 DIM on both color and grayscale images for 3500 steps. Useful for generating pinup images, and the training images were largely sourced from the official manga as well as pinup images from Synecdoche and other fan artists. Given the source material, expect sensitive generations unless you prompt otherwise, but it should be able to generate SFW material as well. The model is robust enough to be used in combination with style LORAs.For best results, include "dark-skinned female", "red eyes", and "pointy ears". As the dataset included both color and grayscale images, be sure to specify one or the other in the prompt or you may get partially colored images. I'm looking forward to the anime, and I'll likely train a second version that includes shots from it in the dataset when it releases. If you have any feedback or would like to get the dataset for yourself, please contact me on Civitai.
