Very like vn backgrounds LoRA, but the dataset is synthetic.
The base model is Inazuma // 稲妻
niji_vn_bg.safetensors