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
Description
first version