This is a first serious try at Lora training.
It is trained on AI generated images.

You can obtain fair result with character only just by using "Mozag" tag, however, in more complicated pics, you might need to use brackets and/or add additionnal tags such as "armor", "leotard", "redhead" etc...