Since the model is trained exclusively on images from the movie, will not perform as well in other situations.

LORA trained on 50 screencaps of Cameron Diaz in the Mask.

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

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.