This LoRA is intended to replicate the likeness of the character American Dream (Shannon Carter) from Marvel Comics. It was trained on 80-ish images, mostly comic panels. I've had pretty decent success with 0.9-1.0 weights, but your success may vary depending on what sort of checkpoint you use. Comic/anime models tend to have an easier time sorting out the details, but it still looks pretty good on most real/semireal models that I've tried.

Trigger word: AmericanDream,

Appearance helpers: blonde_hair, long_hair, blue_eyes, red_lipstick, mask,

To show Shannon's face without her mask, remove 'mask' from the prompt and add: ADUnmasked

(Some models may be stubborn about keeping the "A" and/or the forehead portion of the mask. In this case, try adding "mask" to the negative prompt, along with terms like forehead_mark, facial_mark, headband, face_tattoo, facepaint, and/or forehead_jewel. Also consider decreasing the LoRA weight.)

Outfit helpers: superhero, bodysuit, star (symbol), red_fingerless_gloves, white_sleeves, blue_shoulders,

Shield: ADShield, shield,

(The shield is exceedingly hit and miss, and if you want to just avoid it altogether, try adding it to your negative prompt.)

Since this was trained on comic panels, there's a good chance you'll end up with speech bubbles in your prompts. To avoid this, simply include the following in your negative prompt: speech bubble, English text,

I haven't figured out how to make the little wings on her head generate with any regularity; adding "head_wings" or anything similar usually just ends up making a huge set of wings sprout from her back. Some models mistake the wings for horns, so you may need to include "horns" in your negative prompt to get rid of them.