Less bad quality images. Different approach on image tags. More training time.
Generated images now more rarely get neon line and black background by default - it was a mistake. Softer style application, more versatile for users to change strength of lora.
I love to use lora with additional tags like: oekaki, jaggy lines, thick outlines, thick lineart to get more real digital messy drawn style, less of ai feeling.
Better try yourself to find the best settings and prompts. Creativity is everything here!