Character model based on 34 (pony) / 38 (noobAI) images of Indiana from gacha game Azur Lane. Namely the official splash arts, a few official artworks, some screenshots from the music video, and the small handful of fan arts.<h3 id="usage-42dhxgzii">Usage</h3>BasicWorks well anywhere between a strength of <code>0.8</code> to <code>1.0</code>.Base trigger is <code>indiana</code> Specifying <code>dark skin</code> shouldn't be necessary, but may help improve the result.The <code>face scar</code> is tagged separately in case you don't want to include it, but can be inconsistent.Higher strengths (<code>0.9-1.0</code>) will more accurately capture the <code>face scar</code> but may lead to the base outfits bleeding into the output.OutfitsOutfit components are tagged separately so you can choose what to include.<ul><li>Base skin: <code>long coat, fur coat, necktie between breasts, harness, cleavage, midriff, black chaps, high-waisted thong, black boots</code></li><li>'High Noon Arrest' skin: <code>cowboy hat, blue cape, tassels, black bra, cleavage, under boob, belt, hotpants, cutoffs, chaps, crotchless legwear, fingerless gloves, brown boots</code></li></ul><h3 id="extra-m49b4s3f1">Extra</h3>In the interest of openness and community-helping, I have uploaded the dataset along with the model, this includes: images, captions, and the training json (for kohya_ss).

Indiana (Azur Lane) [Character] (Pony / NoobAI)

<ul><li>Increased dataset size to 34 images (+12)</li><li>Upscaled the lower resolution images</li></ul>Wider dataset means overall style is a lot less affected now, and the upscaled images means that the face is reproduced a lot better. <code>face scar</code> consistency is a little improved, but not much.

Files with SHA256: b9caa40766b55517956ee98fdc09245e4643406220e1ccdac44acded4fffbbd1