This is realistic LORA version of Vocaloid Lily of aniM.O.V.E. Uses datasets of existing cosplay photos for training. Compatible with most of PonyXL-based models. Requires extensive prompt keywords (See how-to and also image samples for example)

How to use this LORA:

1. Use the keyword vocaloid riri to activate this LORA.

2. This lora requires a very extensive keyword in order to generate it as accurate as possible, see the activation tags for further details.

3. Use ((keyword)) tags to forcing to include or omit the object or characteristics from the image results.

4. Set the cross-attention optimization to scale-dot print (SDP).

5. Set the clip skip to 2.

6. Use Adetailer to fix blurry or uneven images when generating with higher resolutions.

7. Usage may vary between models.