Finally, my first Lora is complete. After many failures and retries, I made some major
changes to the trigger words. Please see the details here for how to use them, or you can directly view my image.
Version 1.2 features further fine-tuning of the trigger words and introduces a new "Full Coverage" stage. During testing, I discovered that this update enables a range of additional effects—such as rendering genitalia beneath clothing or adding an organic, biological texture to garments. I utilized higher parameter settings during training to ensure that all elements remain as controllable as possible. So go ahead—unleash your creativity and show me just what kind of spectacular results this model can produce!
However, too much data can make the image look cluttered; for a cleaner portrait, try V0.7.
Basic prompts for portrait：ff000,close-up, portrait, slime suit, glossy skin, viscous, tendrils, veins on face, organic armor, dripping, covered neck, tentacles on face, tentacles in mouth, tentacles in nose, live clothing, [color] substance, [color] bodysuit
1. hh101,covered jaw,
2. hh102,covered mouth, mouth mask,
3. hh100,live helmet,short hair,
4. hh103,live helmet, blindfold, no eyes, covered eyes, no hair
Basic prompts for full-body images：hf101, full body, glowing veins, tendrils, slime suit, glossy skin, viscosities, veins on face, organic armor, dripping, covered neck, tentacles on face, tentacles in nose, tentacles in mouth, live clothing, [color] substance, [color] bodysuit, glowing [color2] lines,
1. fh101,covered jaw,
2. fh102,,covered mouth,mouth mask,
3. fh100,Living Helmet,Mask open,short hair,
4. fh103,blindfold,no eyes,covered eyes,no hair,
hf351, no eyes, covered eyes, no hair, long tongue, sharp teeth, pointed tongue, eyes on chest, monster girl, completely covered, parasitic, tongue out, chest pattern
other
futanari,covered penis,covered testicles,
If necessary, I can optimize the dataset to generate a cleaner, more streamlined version.
Thank you for your interest in my humble work.
Description
Finally, my first LORA runs well under normal circumstances, but it's a bit large because I trained it with large parameters in pursuit of accuracy (which actually didn't work well), so the file size is also large.