Character model based on <s>10</s> <s>15</s> <s>31</s> 43 images of Duisburg from gacha game Azur Lane. Very good reproducibility of the character and outfits now since v3.<h3 id="usage-rbt573l84">Usage</h3>BasicWorks best at a strength of <code>1.0</code> if you want accurate base outfits, turn it down to <code>0.8</code> if you are getting outfit bleed. Base trigger is <code>duisburg</code>OutfitsOutfit components are tagged separately so you can choose what to include.<ul><li>Base skin: <code>black sweater, turtleneck, black jacket, red belt, open jacket, cleavage cutout, underboob, black thighhighs, thigh strap, gloves, manjuu, black boots</code></li><li>Endurance Training (Gone Wrong) (not very accurately reproducible on v1.0): <code>rabbit ears, hairclip, choker, black lingerie, lace-topped thighhighs, lace-topped panties, lace bra, between breasts, detached sleeves</code></li></ul>IssuesAfter much frustration at strange outputs on some seeds with versions 1 through 3, I think I finally figured out what's been happening.The most consistent things coming through were a Blue Archive-esque halo, and black hair, sometimes with a pink underside:<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7b76263a-05ee-43b7-bea1-927a3dc7aeb5/width=525/7b76263a-05ee-43b7-bea1-927a3dc7aeb5.jpeg" />After scrolling through some BA characters, I finally found the culprit: <a target="_blank" rel="ugc" href="https://bluearchive.fandom.com/wiki/Kyouyama_Kazusa">Kazusa</a>My guess is that Blue Archive data is so massively overrepresented on Danbooru that it skews the output of NoobAI towards those characters, i.e. it saw Duisburg's pink-ish eyes, short hair and red streak and thought "that must be Kazusa".In v4.0 I have attempted to mitigate this by reducing the training rate a little and training for more steps, 'overtraining' a little bit to hopefully make sure the model's base info is overriden by Duisburg's appearance. It seems to work a lot more consistently now, but I did still have an ~1/40 failure rate where the output was infected by what I'm calling the "Blue Archive Plague".<h3 id="extra-ksjqivk0f">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).

Duisburg (Azur Lane) [Character] (NoobAI)

<ul><li>trained on 15 (+5) images on NoobAI ε-pred 1.0</li><li>much improved reproducibility of lingerie skin</li></ul>

Files with SHA256: 2714527be1afaf30822ce63fd843e485ec7beb6d5ee37d93fa98b7b7924cc704