The base model only knows flat aviator-style sunglasses, so I decided to train a LORA, which can give your character a sleek wrap-around sporty sunglasses with various colors of frame and lenses. May not work in some cases, so you can combine 3 different versions I made with same dataset, but different settings.