LORA trained on 70 publicly available images of Sarah Snook - mostly images of Shiv Roy from Succession.

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.