This LORA model is trained with majicMIX realistic. it works pretty well with any photorealistic models.

Recommend:

• Image Size: 512x512 or 512x768

• Steps: 20-30

• Sampler: DPM++ 2M Karras

• CFG scale: 7-9

• Weight: 0.85

• Prompt: lips, looking at viewer, upper body

• Expression or Accessory: smile, glasses, round eyewear

Trigger Word: mirachi.s