Female tongues and mouths
Model intended to generate more realistic tongues and mouths.
Can produce mouths and tongues of any shapes or sizes, photorealistic if desired. Given the appropriate prompt, long tongues from likely human to anything.
Trained on SD 1.5, tested with models Deliberate and Dreamshaper
Provided samples include base SD 1.5 model, Deliberate and Dreamshaper. These are RAW samples that have not been improved with GFPGAN or CodeFormer. Provided samples of realistic output, fantasy, humongous, etc.
Triggers: tongue, tongue out, mouth open
More samples on https://www.deviantart.com/znort2005
Description
Features:
Reduced the number of unwanted artifacts
Improved consistency
Training:
2k selected images
20k steps, batch size 4
Learning rate 5e-5:2000, 5e-6:10000, 5e-7:20000
Hypernet structure 1,4,1
Hardware AWS R5g.xlarge
Duration 24h
FAQ
Comments (13)
Much appreciated, this has been a problem for a while.
Thanks for the supportive comment. Don't hesitate to post anything satisfying you generated with it ;)
No safetensor?
Hypernetwokrs only comes in PickleTensor, can't convert it to SafeTensor
Do you just install this into the lora folder is is it something else?
Simply drop it in "stable-diffusion-webui/models/hypernetworks" directory and add <hypernet:tongues-and-mouths-v2:1> to your prompt. To trigger it add "tongue", "tongue out", "mouth open", etc. to your prompt aswell.
I wish I could get this to work. I've done everything correct, and placed it in the right folder. The only thing I'm getting is the model with her mouth open but no tongue coming out whatsoever.
It works only with Stable Diffusion v1.5 model and derivatives. On the other hand, do you use Stable Diffustion Webui ? When placed in the stable-diffusion-webui/models/hypernetworks" directory and add <hypernet:tongues-and-mouths-v2:1> to the end of your prompt, then also and "tongue out" to trigger it this way. Could be something in your negatvie prompt aswell.
You can try this on RealisticVision model as an example:
Prompt: Portrait of a beautiful slim fit woman with her tongue out, white skin caucasian, mouth open, smile, (long tongue), eyes open, looking a the camera, brown hair, short hair, tank top, very detailed <hypernet:tongues-and-mouths-v2:1>
Negative prompt: ugly, morbid, extra fingers, poorly drawn hands, mutation, blurry, extra limbs, gross proportions, missing arms, mutated hands, long neck, duplicate, mutilated, mutilated hands, poorly drawn face, deformed, bad anatomy, cloned face, malformed limbs, missing legs, too many fingers
Just let me know if you still have any issues.
The issue is the files downloads as "femaleTonguesAnd_femaleTonguesAnd.pt" but the hypernet tag <hypernet:tongues-and-mouths-v2:1> needs to match the filename. Rename the file in your hypernetworks folder to tongues-and-mouths-v2.pt and try again. @znort you probably need to change the file that is for download to tongues-and-mouths-v2.pt
@sdu Thanks great find, I wouldn't have noticed that the original filename have changed after the upload. I still use the original local file. I'll be cautious when I'll upload updates or new models.
@znort It's ok I got it to work. My only issue is sometimes you can see a faint outline of the bottom lip through the tongue.
it seems that there is no saliva effect. You can consider upgrading the version to include saliva effect
can we get more triggers? like tongue up? bottom of tongue? licking? touching upper lip? tongue piercing? seems like stable diffusion just doesn't understand those words.