Trained on 117 screenshots of them from the anime. It is heavily biased because I manually picked the images from the anime.
mio.safetensors