A lora with blue hair, using it makes the images more realistic compared to those directly generated by Flux. Is this an additional feature? I use the Nunchaku quantitative model to generate tests. You can use the full-blooded model for testing. I'm very much looking forward to seeing the corresponding image effects