Having trouble getting the AI to figure out what direction you want your character in? Me too.
Designed this Lora to recognize simple standard poses to help guide it to what you want a little easier.
Made with 5 views: front, back, profile (side), three-quarters, and three-quarters back.
Struggles a little differentiating between 3/4 and 3/4 back, but a little push here and there guiding it should take care of it fine.