Trained on a large batch of Daz3d Images then further finetuned at 2k on a small subset for high frequency detail
3D_Style.safetensors