Josie Canesco Instagram Model

NOTE: LoRA is <lora:JosieCanesco:1> in my tests i used the epoch, please remmeber to use the right LoRA name after copy/paste the prompt !

Feel free to experiment the settings you wish but I'm going in any case to drop a few paramters I have already tested that gave me the best results.

Sampling Methos: I strongly recommend to stick to the SDE related or Euler a. Anyway DPM++ SDE Karras worked always out with the best results

Sampling steps: everything between 20-60

CFG Scale: everything between 5-9

Hires Fix: 4x UltraSharp / 4x NMKD Superscale SP 178000 G

I have generated a few images just to better help you understand the quality as well the flexibility. Enjoy

I am always open to suggestions / helpto improve the LoRA !