This is the SD 1.5 version. Best results with suggested settings below.
Note: LoRAs work to different degrees on different checkpoints. Don't leave a bad review because you used a random checkpoint that was overtrained on another face and consequently mucked up the output. Constructive feedback, on the other hand is always welcome.
Suggested settings below:
CFG Scale: 6-7.5
Steps: 30-60
Strength: 0.8-1
Samplers: DPM++ 2S Karras, Euler a
Hires fix using the 4x-UltraMix_Balanced upscaler
ADetailer