GingerV1

This model is designed to be an attractive option to the "overtly model-esque" typical Barbie-doll like models for most female characters being created. The original images used for the training dataset were created using Perchance.org and then merged with other generated images until a set of 35 viable images were ready. The LoRa was trained using Prodigy and Flux and a higher ratio of the generated face/head images to lock in more detail.