Kissing Foot (Yuri) -SDXL

Details

Model description

Recommended Conditions

Trigger:

kissing foot

+ either of:

reclining while knee/leg up, sitting while knee/leg up,

on back while knee/leg up, standing while knee up, on side while leg up (optional)

Strength:

1.0

Positive Tags:

2girls + expression tags

Negative Tags:

kiss (optional)

Description

This LoRA assists in generating a situation where one woman kisses another woman's foot.

As a certain pioneer creator commented, this concept contains several difficulties and needs your prompt arts and luck. Too simple prompts such as "2girls, kissing foot" often summon two girls both kissing somebody’s foot.

Although not trained at all, “licking foot” can also work (the hit rate might drop).

To be exact, “kissing foot” is not a trigger tag but virtually serves as it. The posing tags above (e.g. reclining while knee up) are used as triggers instead because it was difficult to embed multiple poses in one trigger. Nevertheless, you can write prompts without using them.

The triggers above are sorted by their obedience. “Knee up” and “leg up” are both booru tags, but they include much ambiguity. I classified reference images along with their definitions as much as possible, but its effect is questionable.

Because two casts have very different eye-level, it is recommended to weaken major pose tags (e.g. reclining while knee up:0.6-0.8), or you will often see an image of two girls aligning side-by-side and posing as specified.

When you want to get a kissing girl to face toward the backdrop of an image, use “from behind” with weakened strength.

This LoRA was trained all with YURI (= lesbian) images because of my preference. However, it could work also with straight couples.

Sample images

All the images were created with Euler a + Karras on ComfyUI.

Tips

When you very often get images in which the kissed person also tries to kiss something or two persons both pose in the same way, please try: 1) weakening the action prompts, 2) adding expression prompts and 3) not using LCM LoRA. These are common techniques to generate images of multiple persons acting differently.

Quality tags might reduce the number of available poses in your model. Please try removing them if you are not satisfied with your images.

Images made by this model

No Images Found.