This is the LOCON counterpart of the <a target="_blank" rel="ugc" href="https://civarchive.com/models/161053?modelVersionId=181235">AdamW and Prodigy Pantyless Garterbelt LORAS</a>Well... I was getting frustrated as my model kept adding panties when i prompt a garter belt. So in a fit of righteous rage I trained a model for that exact purpose. I gathered the dataset, cleaned it up and oh surprise! Most images are censored frontal shots... I did what i could but the dataset is what it is. What is the meaning of a pantyless garter belt if there's no pussy to see? So you'll either get a small slit or nothing, pubic hair is toggable, but we really need a good frontal pussy LORA. I really hate training concepts because the goalpost is never static. So I trained this and it does what is says on the can for what is worth. And yet from time to time the damn panties keep appearing despite all...More complains incoming. SD knows the concepts So I had to manage it with a feathery touch least it get overcook. The loras are a Dim 1 alpha 1 at 8 epoch 284 unique images with 1 repetition.I did several bakes resulting in two pairs LORA and LOCON in AdamW and Prodigy respectively. The prodigy version is more consistent but is crap for mixing with character LORAs, the adamW is less consistent but mixes better. I will add the 4 versions, 2 here separated by type.For AdamW without a character LORA, it seems to work ok from .8 upwards. All sample images were made with weight 1.Try adding to your prompt: &lt;lora:Garter_belt_no_pantiesV4:1&gt; garter_belt_thighhighs_pussy_no_panties,For Prodigy without a character LORA, it seems to work ok from .7 upwards. All sample images were made with weight 1.Try adding to your prompt: &lt;lora:Garter_belt_no_pantiesV6:1&gt; garter_belt_thighhighs_pussy_no_panties,For AdamW with a character LORA, it seems to work ok from .7 downwards.For Prodigy with a character LORA, it seems to work ok from .6 downwards.Anyway, I'll add the dataset. Don't expect too much unless a tag has a huge appearance rate. Remember this was trained with a bigger dataset with lower repetitions.

Files with SHA256: 16075ac684bc554a4c91df7500fcc632ab605b8153a5c4ca0dc82a41af4e1b76