This model helps with generating images of characters not looking at you, so that you can look at their back without looking at their face as well.