Nävis - Wake / Sillage - Philippe Buchet

Usage: Recommended strength between 0.8 and 1.0.

Tested on RDBT Anima, results may vary with other base models. Avoid using it with base animapreview3.

The LoRA is very stable and responds well to other artists' styles. Sample prompt and generation parameters are available on the first image.

Known issue: Like most character LoRAs, placing two characters in the same image can cause characteristic bleed. To minimize this, explicitly prompt each character's distinct features, even if the base model would normally handle them automatically.

Generating text can be difficult at 1 str, lower it until it works.

There is tons of different hairstyles and clothes in the dataset, you should usually prompt a specific one if you want consistency.