Intended for realistic, sexually suggestive tongue gestures - licking lips or tongue on upper teeth. But, these expressions need not be sexual - they are all over in mainstream ads for toothpaste!
Works best with face close-ups. Avoid wide shots where the face is small in the frame.
You'll see many example prompts have "close-up" or even "extreme close-up"
I haven't tried it with a face detailer, but suspect it would work.
Start from the example images. This LoRA is twitchy and can fail hard. Then, keep trying and tweaking
Example prompt: "Face Close-up of seductive woman, sxxxymouth, parted lips, tongue on teeth," -- then add your details.
Do use the trigger word: "sxxxymouth"
Do not just end your prompt with "sxxxymouth" - it will not work
Using trigger words early in the prompt, near face description, is key
Add one or more of the following phrases as well:
One of: seductive smile, sexy grin, sexy smile, sly smile
One of: open mouth, parted lips
One of: tongue on teeth, tongue on upper lip
NOTE: "Licking her lips" ironically sometimes fails - better to use "tongue on upper lip"
Experiment and have patience. Start with just a couple words, and add from there. For example, adding just ",teeth," or ",parted lips," can be enough to coax a better image out of it.
Trained on FLUX.1 dev on ComfyUI.
Trained on real faces, but works with anime styles too.
Most examples generated locally with ComfyUI, but confirmed at least v1.1 works on CivitAI.
It usually works well at strength = 1. If the effect is too strong (e.g. tongue is too large), reduce strength to .8-.9 or lower. Wider angles often need higher values, e.g. 1.10 or more.
Trained on women only, but see example showing it works for men, and even a 3d bear.
Other LoRA's that affect faces or expressions may conflict or block the effect.
To my delight, can work with 2 people - giving expression to one or both.
Flux only for now. I finally released this version after a couple months of trying to get a LORA for SDXL and Pony to work. If anyone is really interested, let me know and I'll see if I can get a version I'm happy with.
Message me!
I want to help you make it work :) So, msg me! I'm happy to figure out how to get the image you want.
Post your images here!
ALSO - highly recommend you start with my example images and prompts, and tweak to your liking.
This is my first LoRA. Feedback appreciated!
THANKS to @Tenofas for his FLUX LoRA trainer on ComfyUI https://civitai.com/models/713258/flux-lora-trainer-on-comfyui
Trained on a 4070 Ti SUPER with 16GB VRAM.
Have fun!
I_Am_LC
Removed full face images from training data for LoRA, which were affecting other areas of the face (eyebrows in particular). This version is even more tricky to use, but when it works, adapts to a much wider range of faces/poses. As with v1.0, check out my examples.
You will still get the best results with face close-ups.
While Euler/Beta is a good start, try out different samplers and schedulers. See my example images.
Also, with this version, strengths under 1.0 work better in some situations.