Trained with 103 images. The samples were generated with PonyRealism V22, AutismMix.
Recommended settings:
DPM++ 2M, 25-30 steps
Lora weight: 0.7-1
Basic prompt: 1girl, carnival outfit, feathers, jewelry, hair ornament
The trigger is "carnival outfit", you can mix and match the rest of the tags as you see fit.
You can change the color of the outfit by prompting for different colored feathers.
Higher lora weight will make the outfit more detailed but it can make the person's face distorted. I recommend using highres fix.
Put "crowd" in the negative prompt if you want a solo picture.