<h1>V3 trained on NAi prun tags try using strength 0.5-1.0 images was trained with simple background. Using 0.7 and above need location description or simple white background, the tags are skirt, miniskirt ,microskirt using microskirt seems to get more school uniform like images if outfit is not something else. <strong><em>Not tested</em></strong> with other lora</h1><h1><strong><em>V2 has problems with following prompt i reccomend useing along with character loras at lower strengths <s>or random gens with less decrpions</s></em></strong></h1><h3>Mostly tested at strength of 1.0, redone with images from old version and previous attempts. images have more body range when body not described than last version, u have to get creative to get smaller body types petit seem to be hit and miss <em><s>miss mostly</s></em>. more indoor images when location not described to last version's mostly outdoor.</h3><p></p><h2><strong><u>Trained on AOM3A1</u> <em>previous versions trained</em> <em>on</em> <em>AbyssOrangeMix2_hard</em></strong></h2>

ConceptShortSkirt(for extra shortness)

<p>Trained with 6 images on AbyssOrangeMix2_hard</p>

Files with SHA256: 6cadff41fd8724c18285715633cb988a1a114d2f425567ae85c208cb38267b0f