Character LoRA for Sylpha with embedded style (@ninoko).

I was surprised this didn't seem to exist. V0 is an early test draft with only a slight refinement to the dataset and tags. V1 has significant edits made to the dataset and is much higher quality. V0 is a little overtrained, so consider turning weight down as needed. Currently struggling to get Civit to read metadata with my images as an update to my comfyui seems to have broken my save workflow.