<p>A very attractive female Trucker/YouTuber from Europe. Model was trained against the sd_xl_base_1.0_0.9vae model. All of my sample images were generated using ComfyUI along with my recently published RsmGalaxyRealCartoonXL_V1 model. So your results may vary if using the model in a1111. Also, model was trained using kohya_ss as a lora locon. The model seems to often generate decent likenesses straight out of the box. However, I recommend trying 1) An initial gen of an image. 2) Use ClipSeg (ComfyUI) to isolate her face with parms "female face" with blur 7.0 threshold of 0.3 and dilation of 10. 3) Cut By Mask. 4) Hires Upscale of the crop to 1024. 5) Additional sampling on the face crop. 6) Rescale that image back to the original crop size. 7) Paste By Mask onto the originally generated image. I most often get outstanding results using this methodology. Just a suggestion. </p>

Files with SHA256: 5c2e033dc5ee83d9df023e4d8d00e58fbeeb78134e5144f7dc0d3a3af733614a