(V2)

This version features a more pronounced style, with improved detail consistency and better hand and finger rendering.

This LoRA was trained on a dataset of works by the artist pokkora.
The trigger word is: pokkora

Due to the limited amount of high-quality images in the dataset, I strongly recommend using an upscaler (at least ×1.5) along with ADetailer for the best results.

Recommended strength: 0.6 – 0.8
Going above 0.8 may result in a noticeable drop in quality.

Please note: this LoRA may occasionally struggle with fingers, so adapt your negative prompt accordingly.