19 images, 5000 steps. Works well on lots of models but best on Unstable Illusion and Realistic Vision

Training images I think were too close up so it has a huge bias towards white clothing. Need a better dataset if anyone has one.