This character is built around an imaginary female figure, developed from a random collection of images I already had lying around. At some point I decided to expand on them and create a LORA, purely for fun and questionable science. Since this is my first attempt at making a LORA, it’s probably good at nothing and bad at everything. I made it for my own amusement… at least until I discover just how painfully terrible it actually is.

[1.1 update] I don't know why this replaced my 1.0 version, but anyways, this dataset is cleaned up and moderately optimized. Most of the low‑quality material has been removed, and some new questionable samples inevitably slipped in because that is how datasets behave. Overall performance appears improved, or at least it convincingly pretends to be. No keyword required. Strength around 0.75 seems stable, although I have not performed anything resembling a scientific evaluation. Your mileage may vary, especially if you enjoy adjusting settings that probably should not be touched.

Anyway, here she is: Samantha – the neighbour you never had. A 20‑something blonde, perfectly average in every possible way.