LoRA trained on images of characters inviting the viewer under bed covers/blanket. Trained mostly on furry art, but it seems to work fine with humans, too.

Taking into consideration I had no idea what I was doing, it went pretty well.