LORA trained on 70 publicly available images of Naomi Kyle.
Trained on base SD 1.5, should work with most realistic models.
Use ADetailer if details are lost and strength of 1 should work on most prompts.