Trained solely on 4K screenshots from Enchanted
Added "close up" captions to improve finer details on the face