<h1 id="neko_plus_plus-v1.0-pdxl-zayvn1byi">neko_plus_plus <span style="color:#fd7e14">v1.0 - PDXL</span></h1><h2 id="tldr-zgvgjz2ae">tldr;</h2><h3 id="this-is-a-model-for-making-realistic-porn-featuring-human-women-with-animal-ears-a-tail-and-clawspaws.-uaskt1wa2">This is a model for making realistic porn featuring human women with animal ears, a tail, and claws/paws.</h3><p>Positive Prefix: "score_9, score_8_up, score_7_up, BREAK, photograph, realistic, cinematic, BREAK"</p><p>Negative Prefix: "score_4, score_5, score_6, BREAK, text, logo, watermark, bad hands, bad feet"</p><p>Steps: 28-42 CFG: 7-9</p><p>Sampler: euler a (fastest), dpmpp_2m (preferred), heun (best &amp; slow)</p><p>Scheduler: normal, karras (preferred), sgm uniform, beta</p><p>Resolutions: 1024x1024, 1216x832, 832x1216</p><h2 id="animal-prefixes-48lgp1yjv">animal prefixes</h2><p>Avoid directly prompting "catgirl," "doggirl," etc, for best results as these will result in a style closer to furry. Use one of the following immediately after the positive prefix. Other trigger words include "collar" and "fangs." Note that this model should create a result without human ears more often than not, but prompting for long or curly hair will reinforce this behavior, while prompting for short hair is not reccomended and will usually introduce the need for inpainting.</p><ul><li><p>Nekomimi: "cat ears, tail, (paws), claws"</p></li><li><p>Doggirl: "dog ears, tail, fangs, (paws), claws"</p></li><li><p>Kitsune: "fox ears, tail, paws"</p></li><li><p>Deergirl: "deer ears, spotted skin, hooves"</p></li><li><p>Mousegirl: "mouse ears, tail, paws, claws"</p></li><li><p>Bunnygirl: "bunny ears, (paws), fluffy tail"</p></li></ul><h2 id="purpose-hvzn2wzsq">purpose</h2><p>I'm a long time fan of the nekomimi/catgirl style which features animal ears, a tail, and sometimes not quite human feet and hands, which I encountered in visual novels. I've struggled to find a model/LoRA combo that produces that specific aesthetic well in a realistic style without regularly generating a furry style instead. Thus, I created this checkpoint. This is a merge that incorporates a LoRA which was trained on randomly prompted images generated by another specialized merge which were inpainted to reinforce elements such as a lack of visible human ears. <span style="color:rgb(193, 194, 197)">The prompt generator I wrote is called PromptByMood and can be found at: </span><a target="_blank" rel="ugc" href="https://github.com/RLW-Chars/comfyui-promptbymood"><span style="color:rgb(193, 194, 197)">https://github.com/RLW-Chars/comfyui-promptbymood</span></a></p>

Files with SHA256: b456f9782b99e1b2765142035499501cfbd52594bd69b1b03fd2982075cf6215