Juvia Lockser from Fairy Tail — character LoRA for Illustrious / SDXL anime models.

Trained on a curated dataset focused on face consistency, body shape and her signature outfits. Epoch 4 checkpoint — clean face, stable across angles.

Trigger word: juviaWAI

Recommended weight: 0.6–0.8

Best with: WAI-NSFW-illustrious-SDXL or any Illustrious-based checkpoint.

Tips:

- Add "masterpiece, best quality" to the prompt — the LoRA learned those tags from the captions.

- For a cleaner face keep the weight at 0.7 and add "detailed face, looking at viewer".

- Stacks well with glossy / 2.5D style LoRAs on top for a semi-real look.

- Water magic works best with "water, water splash, water droplets" in the prompt.

Good at: portraits, full body, canon outfit, casual clothes and swimwear, NSFW.

Weak at: extreme close-ups where no face is visible, heavy armor.