hey, i want tell you about my lora
https://civitai.com/models/119376?modelVersionId=142718
its trained on 300 images of old and young woman in close and far distances with and with out smile on 11 different cultures
and it works very well with the SDXL_base_model
first impression
with my lora, 25steps
...
without my lora, 25 steps
...
now a step by step image generation with more weight in my lora
without my lora, 42 steps
...
with my lora 0.5, 42 steps
...
with my lora 0.7, 42 steps
...
with my lora 1.0, 42 steps
...
you see it change the picture not too much, but you get better faces and like you see, ore with less steps . . .
THY a lot for merge and inform me PLS :)