A character LoRA for Anarizand.

This version was trained with multiple outfits, hairstyles, viewing angles, compositions, and background variations in order to reduce overfitting to a single costume or background.

The main goal is to preserve the character's identity while keeping outfits, hairstyles, poses, and backgrounds relatively flexible.

Trigger word:
anarizand

Core features such as white hair with subtle pink accents, red eyes, and red horns were mainly learned as part of the character identity.

The training dataset includes:

• Front, side, back, and three-quarter views

• Full-body and upper-body compositions

• Long hair, ponytail, and low twin tails

• Gothic dress

• Sportswear

• Winter coat

• Knitwear and pleated skirt

• Pajamas

• Multiple simple and gradient backgrounds

For better clothing control, explicitly describe the outfit you want in the prompt.
The LoRA can also be used with detailed backgrounds such as forests, interiors, and outdoor scenes.

Primarily recommended for single-character generation.
When multiple character LoRAs are used together, some features such as hair color or horns may occasionally bleed between characters.