LORA trained on 37 publicly available images of Lea Seydoux.
Trained on base SD 1.5, should work with most models.

Use ADetailer if details are lost and strength of 0.75 should work on most prompts. Use lower strength for other models.

(21, 16 images) - (24, 18 repeats). Picked the 8th epoch.
Consider leaving a review if you like the model. All feedback is appreciated.