This is a simple LoRa, trained on 100 MLP-styled pictures, both official and fan-made backgrounds.

If i get to know training better, i will re-make this one, but not in a visible future