Style based on the artist 03_bara

Trained using 136 images scraped from gelbooru

I used this article to train loras in the past. But with PonyXL models I only use the linked Collab Notebooks to produce my datasets then use Civitai's on-site training to train the models themselves.

Dataset was tagged both by the Collab Notebook's tagging step and then since it skipped over some images I retagged everything using Civitai's on-site auto trainer prepending the artist's name to the tags.

First sample images were done during on-site training