There's some video tutorials about how to create a consistent character (Called "Lora" or "Textual Inversion") for Stable Diffusion a YouTube search will give you quite a few. You also might want to look into ControlNet for posing. That can be helpful to get a consistent composition at different sizes with different outfits.
I got automatic1111 installed and made a bunch of pics with a few models. They look great, and the character is pretty consistent. But I cannot figure out Loras for the life of me lol. Whenever I try to img2img and throw a Lora in the prompt, I get an error and a full black image. But thanks so much for getting me started on searching for the right things!! I'll hit up their discord for help later tonight.