This LoRA was trained on Fire Emblem: Radiant Dawn (FE10) portraits and does a good job at replicating that art style.

I've found the best way for me to prompt the images is to add booru tags at the beginning for the art style of pixel art, then the natural language model.

Also, the first image in is Selenea from Cycle of the Heavens, a free fan game based on Fire Emblem. Give them a look! :)