Vampire watching you from the woods | PainterlyOldStyle

Making this for my "Show Off Your Lora" challenge.

Unsurprisingly (if you've seen my profile), this lora was trained largely on portraits. It's a bit silly that I even made it, because probably 80% of the images supplied to the model were made with the exact same artist mix, and the other 20% still included at least TWO of the main FOUR artists.

The lora was designed to do content like this, and it works pretty okay for that purpose. In some ways it can be even better than referencing the artists, themselves, because it was trained on a set of my favorite results, so the AI kind of understands what I personally like. But there are many subjects it really can't handle, where using the artists, themselves, would allow SDXL to figure something out.

More open prompts (credit originators):