Notes:

Due to the training data set consisting entirely of photos, the faces were cropped out to prevent the AI training on the faces. As a result, AI might struggle to produce images without helmets being worn. If generating such images, would advise adding "helmet, wearing helmet" with sufficient weightage to negative prompt, and then hair prompts with sufficient weightage to positive prompt

Likewise due to point number 1 above, you can't use "hand out of frame" because AI seems to confuse it with "head out of frame" and crop off the face lol

For some reason, there are two different available designs with the chest symbol, one being wing-shaped and one being M-shaped. Images of both were added into the training data set and captioned with "wing-shaped chest symbol" and "m-shaped chest symbol" accordingly, but doesn't seem like AI figured out what is going on here from my testing LOL Source: https://civitai.com/models/971277?modelVersionId=1087571