Some begging first

If you guys want to do me a favor then upload some non-explicit material that you generated with this model.

Background Info

This LORA was trained using a variety of images of Harley King -- an adult actress. The base-model against which this was trained was the standard SDXL 1.0 base model. I used 40 images, and the result you see here is the 5th epoch of the LORA.

Usage

Use the LORA with a strength of approx. 0.8 and the trigger-word "harleyking".

Showcase

Images shown were generated with CFG scale 6 and Euler a with 30 steps (768x1024). The upscaler was SwinIR with a scale-factor of 1.6875.

Positive prompt

harleyking,high quality,extreme close-up,blonde,blue eyes,woman,shirt <lora:harleyking:0.8>

Negative prompt

bad quality,worst quality,ugly,blurry,tattoos,breasts

Misc

The ADetailer was applied to generate the shown images, but the LORA works quite well when not using the ADetailer. Also, this is my first LORA, I still have much to learn, so take the result with a grain of salt...