A LoRA trained on images of multiple angles(specifically face|ass) of people. The only needed trigger is multiple views. I do recommend prompting both the front and back features.