[PonyXL] Flattening

Makes people as flat as paper! I was surprised this worked without a huge amount of training data, so here it is! I might take another crack at this one later to try to make it more consistent and fix the face shading, but I really like that it works with realistic-ish generations already. I had fun mixing this with inflatable/latex/pooltoy-type loras, and elongation.

Usage

A weight around 1.3 for both the LoRA and the trigger word "flattening" usually works well. Add in "curly", "folded", or "in rollers" at heavy weights (1.2+) to get some different results. I didn't find it difficult to get this to work, but it was difficult to get results I wanted.