Lora trained on 150 images selected out of Instagram and google images of Instagram model, influencer and makeup artist @glambyflo (not particularly high quality) by GPT4-vision API. The resulting 150 images are also completely autonomously captioned by GPT-4 vision API. Lora trained in Colab. Recommended strength as usual.
Use keywords like closeup portrait, half-body portrait, full-body portrait etc. to better control distance of subject from camera or in a painting.