-
Version 1 - Alwayes use: "<lora:airiakizuki_xl_v1:1>" for activate Lora
-
Tested with models:
rhapsodyxl_v10
-
Training lora: max img 1024 x 1024
-
Use: x:768 x y:1024 (excelent results)
-
Use low sampling steps, cfg scale low,
-
DPM++ 3M SDE Karras
UniPC
-
Tag prompts recommended:
airi, blue eyes, blonde hair, oni chichi,
-
Negative Prompt Needed
-