Setting the LoRA weight to 1.5 seems to work better.
It works, but there are still some rough edges.
Beautiful_V2.safetensors
