LORA trained on 65 publicly available images of Natalie Portman. (trying 40 face, 25 body split for this one too).

Trained on base SD 1.5, should work with most realistic models (works best with serenity v2).

Use ADetailer if details are lost and strength of 1 should work on most prompts.
Consider leaving a review if you like the model. All feedback is appreciated.