i do accept requests, tho i only have limited resources and time, add me on discord: @skver<h2 id="anima">anima</h2>i hate anima so much, uncontrollable, but its so clear and high res... trained on the same old Nyanners 2.0 twitter dataset as the illustrious version (193 images) plus 27 more hoodie and gothic frames, so 220 images totaltrigger word: <code>nyanners</code>no need to add <code>pink hair</code>, <code>cat ears</code> or <code>purple eyes</code>, the trigger already brings them (and the cat bow clip). pick an outfit tag and add a few words for the clothes, the outfit tag alone gives generic clothesoutfit tags: <code>nyanners_cami</code>, <code>nyanners_lace</code>, <code>nyanners_hoodie</code>, <code>nyanners_gothic</code><h3 id="versions">versions</h3>same lora, same prompts, pick the one that matches your model- base Anima (28 blocks): <code>nyanners_anima_mix2.safetensors</code>- novaAnimeAM (remapped, 40 blocks): <code>nyanners_anima_mix2-e12_remap40.safetensors</code>. the same weights moved onto the 40 block layout of novaAnimeAM_v6029B. use this one on novaAnimeAM, the plain one loads wrong there because the blocks don't line upboth are bf16, rank 16, about 46 MB. use at weight 1.0<h3 id="settings">settings</h3><code>er_sde</code> / <code>simple</code>, cfg 4-5, 40-50 steps, 1024px wide. cfg 7 with euler_ancestral burns the colors (red hair, pink skin)quality tags in front: <code>masterpiece, very aesthetic, best quality, score_9, score_8, score_7, 1girl, solo,</code> and <code>absurdres, highres</code> at the endnegative:<pre><code>worst quality, low quality, score_1, score_2, score_3, artist name, blurry, jpeg artifacts, chromatic aberration, saturated, neon, royal blue, cyan, hot pink, magenta, text, signature, watermark, black ear tips, black ears, cropped, border, cutout</code></pre>add <code>long hair</code> for long, the default is the short bob. <code>light smile, smile, closed mouth</code> for a gentle smile (without it she looks a bit sleepy)<h3 id="prompts">prompts</h3>cami<pre><code>nyanners, nyanners_cami, black camisole, halterneck, straight neckline, choker, bare shoulders, collarbone, upper body, looking at viewer, white background, simple background, pastel colors</code></pre>add <code>v-neck, plunging neckline</code> to the negative for the flat necklinelace<pre><code>nyanners, nyanners_lace, pink camisole, pink bow, black lace trim, ribbon choker, halterneck, bare shoulders, collarbone, upper body, looking at viewer, white background, simple background, pastel colors</code></pre>hoodie<pre><code>nyanners, nyanners_hoodie, pale blue hoodie, white trim, off shoulder, drawstring, black camisole, halterneck, bare shoulders, collarbone, hood down, upper body, looking at viewer, white background, simple background, pastel colors</code></pre>gothic<pre><code>nyanners, nyanners_gothic, black dress, frills, frilled collar, long sleeves, buttons, black ribbon, rose, hair flower, bat hair ornament, upper body, looking at viewer, white background, simple background, pastel colors</code></pre><h3 id="notes">notes</h3>- mostly head and shoulders data, so full body is weak (only the hoodie had full body frames)- keep <code>pale blue hoodie</code> and <code>hood down</code> in the hoodie prompt, otherwise it goes white or the hood comes up- the cat clip sometimes gets a small extra detail<h1 id="illustrious:">illustrious:</h1>the clothing is awful as usual since i dont really have a good dataset, still using the old Nyanners 2.0 dataset, that i usually use, from twittertrigger word: nyanners(tbh add "long neck" to make it match more the og artstyle i guess)can do both long and short haired versions just for a little help in starting out: gothic lolita: nyanners, 1girl, animal ears, solo, pink hair, cat ears, twintails, hair ornament, hair flower, animal ear fluff, yellow eyes, long hair, ribbon, rose, gothic lolita, fang, bangs normal: nyanners, 1girl, animal ears, purple eyes, pink hair, solo, cat ears, long hair, off shoulder, looking at viewer, bare shoulders, frills, cat girl, blush, animal ear fluff, collarbone, hair ornament, frilled shirt, bangs, upper body, :3, cat hair ornament, ribbon, bow, shirt also has support for her mismatched cat ears thighhighs:asymmetrical legwear, animal ear legwear, cat ear legwear the <a target="_blank" rel="ugc" href="https://static.skver.space/nyanners-dataset.zip">dataset</a> if you wanna cherry pick more stuff from it --------------------------------<h2 id="sdxl:"> sdxl:</h2>same dataset as sd 1.5 modeluse with weight around 0.8-1.0~sample images were generated with animagine v3.1activation:<pre><code>nyanners</code></pre>base prompt:<pre><code>1girl, nyanners, catgirl, cat ears, outfit, hair ornament, pink hair, purple eyes, long hair</code></pre><h2 id="sd-1.5:">sd 1.5:</h2>base lora trained on pictures of nyanners v2 model from twittermerged with <a target="_blank" rel="ugc" href="https://civarchive.com/models/79354/nyanners-vtuber-character-lora">Nyatasha Nyanners v1</a> use with weight at 1~ dont forget to do hiresfix for better results 👍

Nyanners

outfit tags: nyanners_cami, nyanners_lace, nyanners_hoodie, nyanners_gothic28blocks and 40 blocks version (just remapped)trained on 28 blocks anima base

Files with SHA256: 74f845dcdab56725bf114cfd238f9b70a86b2f3f2e1c078ce339ef5f804ab209