Zhen Lu Style Lora trained on/for PonyXL and all its merges.

Dataset has a some manga style images. So, some results with text, monochrome-ness, prevalence of Fate Series characters. Personally, I really like this image: "click" and it was the one that prompted me to try and create the Lora.

This is a second attempt with massively trimmed dataset. The first one had around a 1000 images (most of which were manga/comics), but the resulting generations were very generic, and not really displaying an obvious style. This version has that problem too, but it's more manageable.