Lora focused on waterfalls. Trained on a set of 104 images from around the globe of different styles of waterfalls. This early version isn't super great, as most SDXL checkpoints are focused on people rather than landscapes, but it is a decent starting point. Recommend a moderate CFG (I've been using 13) and 40-60 steps. I've had good luck with several checkpoints including: base SDXL, cinemaxAlphaSDXL, MiaModel, TalmendoxSDXL.

Files with SHA256: e5d6091ef1ce2b28dbf52fe9e63b6564a86ecf65c94baf931c6b7326cff0e7fb