physical violence

weapon violence

wide hips

revealing clothes

thick thighs

downblouse

convenient censoring

huge breasts

pg-13

corpses

suggestive

oral invitation

pg13

sexy

sexual situations

male nudity

disturbing

male swimwear or underwear

female swimwear or underwear

partial nudity

undressed

female nudity

breasts out

exposed female nipple

breast out

lingerie

male underwear

hair over breasts

female swimwear

gigantic breasts

no panties

graphic violence or gore

covered nipples

huge butt

strapless leotard

sitting on face

emaciated bodies

one breast out

nsfw

female underwear

nude

graphic male nudity

adult toys

illustrated explicit nudity

nudity

graphic female nudity

hentai

futanari

porn

sexual intent

genitals

peeing

vore

oral

sexual activity

anal

blowjob

dildo riding

incest

hanging

hate symbols

nazi party

white supremacy

diapers

scat

self injury

hate speech

urine

extremist

child on child

latex clothing

swimwear

bukkake

fellatio

cumshot

implied fellatio

eat_cum

cumdrip

cum in pussy

cum on face

after fellatio

cum on hair

cum on body

cum on tongue

cum on hands

cum in mouth

triple fellatio

autofellatio

fucked silly

cum on pussy

This LoRA is meant to replicate the harness, jockstrap and socks in <a target="_blank" rel="ugc" href="https://coyotejocks.com/collections/varsity">Coyote Jock's Varsity Collection</a>. The LoRA has gone through many iterations. Check the version details on usage.I've trained the LoRA on the following colors, but it can generate other color variations to varying degrees of success:<ul><li>black</li><li>blue</li><li>burgundy</li><li>green</li><li>"neon" black</li><li>white</li></ul>

Happy with how this one turned out. Looking through the versions of the LoRA you can see how it's evolved alongside open-source AI generation.Mentioning the subject's butt and/or nakedness is sometimes needed to generate the jockstrap correctly. Use <code>knee high</code> for the socks to make them longer.Trained on Flux.2 Klein 9B Base with Prodigy and a learning rate of 1 with AI Toolkit by Ostris.

v1.0 [Flux.2 Klein]

Use a weight of 2 or higher when using on ZImageTurbo.Mentioning the subject's butt and/or nakedness is sometimes needed to generate the jockstrap correctly. Use <code>knee high</code> for the socks to make them longer.The Z-Image version also has more issues than the Flux.2 Klein version with generating the jockstrap separately from the harness. Trained on Z-Image with Prodigy and a learning rate of 1 with AI Toolkit by Ostris.

v1.0 [Z-Image]

If the sampler isn't showing in the Flux example images it's most likely the <code>IPNDM</code> sampler.

v1.0 [Flux]

v1.0 [Illustrious]

You can follow the instructions for triggering the LoRA from SD 1.5 v4. The same triggers apply. The base trigger is <code>jock_harness</code>.The LoRA doesn't create the <code>jockstrap</code> or <code>socks</code> on their own (at least to their full strength). If you want those to appear, you must include those trigger words in your prompt. See the model's post for examples.Some notes:<ul><li>When triggering the full outfit you might need to add <code>full body</code> to include the socks as well.</li><li>You might need to add <code>necklace</code> to your negative prompt when generating closeups or headshots.</li></ul>

v1.0 [Pony]

This version incorporates a larger dataset pulled from social media and the ability to generate socks. Hopefully, this helps with the "sameness" of the facesWhat happened to V3 do you wonder? I worked on a new version based on the feedback from V2 and realized that adding a more diverse dataset from social media might help with flexibility. V4 takes into account a larger dataset and V3 was trashed.Use <code>wearing a jock_harness</code> to trigger the LoRA. It can also replicate the jockstrap and (to a point) socks.<ul><li><code>wearing a jock_harness</code> will generate the harness and usually the jockstrap</li><li><code>wearing a jock_harness and jockstrap</code> will give you both the harness and jockstrap</li><li><code>wearing a jock_harness and jockstrap and socks</code> should give you the full outfit</li></ul>

v4.0 [SD 1.5]

I rebuilt the LoRA several times and think it is in decent enough shape now. Fixed a lot of the issues with v1.<ul><li>The LoRA is compatible with more ethnicities like darker-skinned folks</li><li>No need to use <code>cyth</code>, <code>jock</code>, or <code>harness</code></li><li>The likeness of the model in the training dataset is lessened</li><li>The LoRA is more flexible and can be used with other LoRAs</li></ul>Tips<ul><li>Use <code>jock_harness</code> to prompt for the harness and jockstrap.</li><li>The LoRA likes them beefy, muscular, and hairy.</li><li>When prompting for color if you want to replicate the classic or vintage jockstrap look use <code>white and &lt;color&gt;</code>. This should hopefully keep most of the harness and jockstrap white and add colored bands.</li><li>When prompting for a green <code>jock_harness</code> the green color will meld into the person's chest hair...not sure what that's about.</li><li>Sometimes a collar shows up. Add <code>collar</code> to your negatives to avoid this.</li><li>If you're mixing this with other LoRAs use a lower weight.</li></ul>Changes<ul><li>Used BLIP2 captioning - not sure if this helped much</li><li>I pruned a lot more tags in the captions like <code>gym</code>, <code>jock</code>, <code>harness</code></li></ul>

v2.0 [SD 1.5]

This is my first clothing/style LoRA so there are a few caveats:<ul><li>I find using <code>wearing a &lt;style&gt; cyth</code> works most of the time. For example <code>wearing a (blue) cyth</code> or <code>wearing an (icelandic) cyth</code>.</li><li>It's very much tied to muscular and hairy men. It also struggles with darker-skinned colored men, unfortunately. I think these limitations are due to the limited dataset.</li><li>It will often draw the shoulder straps, but not the chest strap. If that happens you can use <code>(harness)</code> to influence the model to draw one.<ul><li>You can use <code>harness</code> and <code>jock</code> to fine-tune the strength of the model. The <code>jock</code> keyword seems to be tied to the face of the male model in the dataset which can limit faces and body styles.</li></ul></li></ul>

v1.0 [SD 1.5]

Jockstrap Harness [Z-Image + Flux.2 Klein + Flux + Illustrious + Pony + SD 1.5]

Clothing