v1 (ILL)

A v2 is required for a more illustrious look since I'm using pony images in the dataset.

Trigger word: 3girls, kiss, tongue

Strength: 0.7 - 1 (Lower the strength for more Illustrious look. Higher strength for more stability but might change the checkpoint style)

v2

Half the lora size of v1, clip skip 2, and changing the dataset.

Also due to my training images use monochrome in negatives to remove it.

Trigger word: 3girls, kiss, tongue

Recommended: anime (for more illustrious look)

Negative Recommended tags: 3d, realistic

Strength: 1

v1

When two isn't enough! The concept is pretty consistent but can sometimes have an extra tongue or an extra girl/head. I've tested with different camera angles like "from above" and "from below" as these work well. "Close-up" is hit or miss. Girls can be sitting, standing, or kneeling. Didn't train for lying/laying.

I notice sometimes that a ???? can sometimes appear between the tongue on tongue action so use ?????, ?????????, 1boy, ????, futanari to remove it almost completely. Ignore this if you want it.

Also due to my training images use monochrome in negatives to remove it.

Trigger word: twk1ck1x, 3girls, kiss, tongue

Quality tags: score_9, score_8_up, score_7_up, score_6_up

Negative tags: 3d, realistic, monochrome, ?????, ?????????, 1boy, ????, futanari

Strength: 0.8 - 1