LoRA trained on 20 fully AI-generated images (no real people used).  
Trigger word: sof.ia  
Recommended weight: 0.8–0.95  
Base model: SD 1.5 / Realistic Vision  
Safe for work & NSFW.