Works with most anime models. Realistic models work, but not very well.
If you like my work, please consider giving me a 5* rating! TRAINED ON 768 RESOLUTION
Best weight: 0.8-0.9 - canvas size in portrait formats work well.
I don't understand why Adina's LoRa with a worse dataset is better than the Nicky LoRA.
Works best with standing images. Because there is no sitting fanart of Adina anywhere I looked...
For best results, include the following tags:
adina, blue eyes, white hair, long hair,
For her original outfit, include:
fated dress, frills, off-shoulder dress, bare shoulders, veil, black gloves, black footwear, holding orb,