Not a internet-famous face, the base model integrates WhiteXL_realisticMix. The training material is derived from the frames extracted from video conversion drawings. I used my refined adjustments with Lora to adjust the overall model texture. Testing at the size of 512*768*2 yields a better effect for half-body images, while for full-body images, it is recommended to generate images at 768*1024*2 to avoid stretching the face. During my testing, I hardly needed to use any face editing tools.