Version 2 smokes version 1.

They are miles apart.

Realistic Vision is great for this.

If you want interracial, just use the keyword in your prompt.

If you don't want it, then just say woman kissing a man, and specify the races/ethnicites.

Here's how you reproduce the images:

1.) Call the LORA with some sort of a prompt along the lines of "woman kissing a black man". Feel free to say what her hands should be doing, or what they should be wearing.

-> This will give you an 80% done image

2.) Then, go to img2img, and kick out the lora, and re-run the model.

-> This will then give you a solid finalized image.