This is part of the CoachBate gear series. A singlet-focused clothing LoRA built from 250+ curated training images covering collegiate wrestling singlets, adult gear singlets, mesh styles, glossy latex/rubber looks, rear-zip designs, open-seat/crotchless variants, and branded athletic cuts inspired by labels like Nasty Pig, CellBlock 13, Locker Gear, Breedwell, Barcode Berlin, Coyote, N2N, and classic sportswear references.No trigger word required. Prompt naturally for the style you want.<h2 id="prompting-guide">Prompting Guide</h2>Use simple descriptive phrases. The captions were written around visible garment categories, views, materials, cuts, and background context, so those same words are useful in prompts.<h3 id="core-singlet-types">Core Singlet Types</h3><ul><li><code>college wrestling singlet</code></li><li><code>classic wrestling singlet</code></li><li><code>athletic singlet</code></li><li><code>varsity contrast singlet</code></li><li><code>low-cut wrestling singlet</code></li><li><code>open-back singlet</code></li><li><code>one-piece athletic singlet</code></li></ul><h3 id="adult-gear-cuts">Adult Gear Cuts</h3><ul><li><code>crotchless singlet</code></li><li><code>open-seat singlet</code></li><li><code>assless singlet</code></li><li><code>backless singlet</code></li><li><code>bareback singlet</code></li><li><code>rear zipper singlet</code></li><li><code>zipper on the ass</code></li><li><code>rear access zipper</code></li><li><code>open rear singlet</code></li><li><code>strappy cutout singlet</code></li></ul><h3 id="materials-and-finish">Materials &amp; Finish</h3><ul><li><code>mesh singlet</code></li><li><code>sheer mesh singlet</code></li><li><code>glossy singlet</code></li><li><code>latex singlet</code></li><li><code>rubber singlet</code></li><li><code>wet-look singlet</code></li><li><code>structured athletic singlet</code></li><li><code>stretch panels</code></li><li><code>contrast trim</code></li><li><code>side paneling</code></li></ul><h3 id="brand-and-logo-prompting">Brand &amp; Logo Prompting</h3>The captions include visible brand/logo language where it appears, including phrases like:<ul><li><code>Nasty Pig bull-pig head emblem</code></li><li><code>NASTYPIG wordmark</code></li><li><code>CellBlock 13 style singlet</code></li><li><code>Locker Gear singlet</code></li><li><code>Barcode Berlin singlet</code></li><li><code>Breedwell singlet</code></li><li><code>Coyote varsity singlet</code></li><li><code>N2N athletic singlet</code></li></ul>The base model already understands major sports brands, so if you want a specific athletic logo or collegiate look, include it directly in your prompt.<h2 id="captioning-notes">Captioning Notes</h2>The dataset captions were designed to describe what is visible without locking the model too hard to one color or one exact product. They include the garment type, view angle, construction style, and setting, such as:<ul><li>front view, rear view, side view, or three-quarter view</li><li>short legs, athletic coverage, open armholes, racerback straps</li><li>mesh, sheer, glossy, latex, rubber, or stretch fabric</li><li>open-seat, crotchless, rear-zip, backless, or strappy cutout construction</li><li>studio background, gym background, locker room, product-style backdrop, or white background</li></ul>This should make the LoRA easier to steer with prompt language instead of needing a rigid trigger word.<h2 id="about-this-preview">About This Preview</h2>This is a preview build trained from a heavily assembled and cleaned dataset with matching captions, garment-focused crops, and multiple product views.I’ve also set up an AI-assisted workflow that watches for new singlet releases, stages eligible images, and prepares caption-ready additions after approval, so the dataset can keep improving as new gear drops.I’m continuing to train and refine the model for stronger logo recognition and cleaner brand-specific detail handling.Preview buyers will receive v1 when it is released, and the model is currently discounted during preview.

Files with SHA256: c10ef2ab0e2d697cd1de7792f9f96c21c74d59fc0a1e33d4984de70b08564ab9