日本語&gt;&gt; <a target="_blank" rel="ugc" href="https://note.com/plum_code/m/m56c4c3dce14e">https://note.com/plum_code/m/m56c4c3dce14e</a> 比較まとめ &gt;&gt; <a target="_blank" rel="ugc" href="https://note.com/plum_code/n/n616a57bda009">https://note.com/plum_code/n/n616a57bda009</a>I´m Japanese. I´m using chatGPT for translation.I have confirmed that my model has been reprinted in various places. As soon as we find out, I will request for the account to be confiscated or deleted. If you have a site you would like to use, please let us know in the comments. Tensor art...Completion PixAI...Completed seaart...model completed, Most of the images on the site remain reprinted.<h2 id="heading-32"><a target="_blank" rel="ugc" href="https://tensor.art/u/625967290114022698">tensor.art </a><a target="_blank" rel="ugc" href="http://pixai.art">pixai.art</a><a target="_blank" rel="ugc" href="https://pixai.art/@plumcode/artworks/models"> </a> <a target="_blank" rel="ugc" href="https://civarchive.com/images/5694661"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/e27114ed-f0c2-45dc-92f9-239fb1f7ea9b/width=525/e27114ed-f0c2-45dc-92f9-239fb1f7ea9b.jpeg" /></a></h2><a target="_blank" rel="ugc" href="https://civarchive.com/models/244733/plumbeer-robustmix"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b5e0fe9f-bb56-4f8e-bbed-ef7f8227d066/width=525/b5e0fe9f-bb56-4f8e-bbed-ef7f8227d066.jpeg" /></a> The model that easily generates male.Someday, I'd like to create Lora, so please give me the my model a unnecessary Buzz. I'm not sure if I'll actually make it, so please send it without expecting much, with a light-hearted feeling＼( 'ω')／ Caffeine, too, please☕ <a target="_blank" rel="ugc" href="https://ko-fi.com/plum_code">https://ko-fi.com/plum_code</a>―――――――――――――― I recommend V1, so please try it once. (´；ω；`) ―V1―――――――――――― Overwhelming scenic details and a fluffy touch. The color palette is subdued, yet it generates remarkably beautiful and intriguing outputs. This model is adjusted for the harmony between characters and landscapes, making it challenging to depict only the upper body of a character. Consequently, there may be distance perception issues with characters. This will turn the character's feet into animal feet if you don't put on shoes. For drawing male characters, I recommend using <a target="_blank" rel="ugc" href="https://civarchive.com/models/244733/plumbeer-robustmix">RobustMIX</a>. ―V2―――――――――――― A petite model fine-tuned with Lora, based on V1. It might be a bit challenging to work with, but it allows for the creation of cuter characters compared to V1. There's a bug in the sense of distance with characters. ―SuprCute v1――――――― V2 has gained more popularity than the masterpiece V1, and I felt a bit frustrated. Therefore, I created an supreme cute and sparkling model. I believe I've optimally incorporated the excellent aspects of V1. Please treat this model with care and enjoy it to the fullest! (*‘∀‘) There's a bug in the sense of distance with characters. PS’ Using (Fluffy body,Fluffy,Fluffy hands:1.5) brings the appearance Ultimate kawaii. ―V3―――――――――――― This model, an improved version over V1, boasts refined visuals in FurryMIX. However, it's less user-friendly than V1, and achieving consistent fluffiness is a bit challenging. Generating animal-like legs, once an advantage in V1, has become somewhat like a gacha in V3. Making characters Anthro can be tricky, but remarkably, there are almost no bugs in character distance perception! Recommended weight (furry,anthro,Fluffy body:1.3),(Fluffy) <h3 id="heading-2">推奨設定／Recommended settings</h3>Steps: 30, CFG scale: 7, Denoising strength: 0.55, Clip skip: 2 VAE: <a target="_blank" rel="ugc" href="https://civarchive.com/models/118561?modelVersionId=131658">Vae-ft-mse-840000-ema-pruned</a>I don't recommend using the elaborated Lora too much, as it tends to make things cluttered(<a target="_blank" rel="ugc" href="https://civarchive.com/models/58390/detail-tweaker-lora-lora">Detail Tweaker,</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/81291/flat2"> frat2</a>). Originally designed to be satisfying with just t2i. (I'm playing with VRAM 6G. i2i takes a long time and causes stress. --lowvram --always-batch-cond-uncond) Sometimes, interesting results can come from simple prompts.<h3 id="heading-199">Prompt: Anthro</h3><pre><code>////quality prompt etc////,(best fluffy hands:0.7),fluffy,(anthro, &lt;&lt;Color&gt;&gt; furry &lt;&lt;gender&gt;&gt;,&lt;&lt;Color animal ears&gt;&gt;,&lt;&lt;1 Color animal tail&gt;&gt;)</code></pre>▲ v3 Recommended weight: (furry,anthro,Fluffy body:1.3),(Fluffy)<pre><code>EasyNegative, EasyNegativeV2,(bad-hands-5:1.2),(worst quality, low quality,human skin,realistic lip,human ears:1.4),error,extra fingers,fewer fingers,bad legs,extra hands,bad anatomy,multiple tails</code></pre><h3 id="heading-199">Prompt: Kemomimi</h3><pre><code>////quality prompt etc////,best hands,(human face,&lt;&lt;Color animal ears&gt;&gt;,&lt;&lt;1 Color animal tail&gt;&gt;)</code></pre><pre><code>EasyNegative, EasyNegativeV2,(bad-hands-5:1.2),(worst quality, low quality,human ears,realistic lip:1.4),error,extra fingers,fewer fingers,extra hands,bad anatomy,multiple tails,</code></pre><h3 id="heading-76">Other recommended Negative</h3><pre><code> telephone poles, electrical wires, ??? street light,Electric lights ???</code></pre><h3 id="heading-87">Ultimate prompt:</h3>Generate immediately without reading the prompts, keeping everything except the number of steps at the default settings!! Please submit the finished image to this model. Step: 20~25 Positive prompt: <s>Best quality, (masterpiece:1.1576), (high resolusion:1.2155), (illustration:1.05), (ultra detailed 8k art:1.05), kawaii, (fluffy), cute, (micro pigs:1) , deformation, multiple Kawaii pigs, lots of micro sized pigs,fantasy</s> Negative prompt: <s>(EasyNegativeV2:0.9), (nsfw, nude:1.5), anime, human hair pig, pig with human face, bad anatomy, bad hands,bad legs, cat, cat ears, cat's whiskers</s> I crafted this with a focus on the Glow in the eyes and the fluffy body, vivid landscape descriptions. I would be delighted if you could imagine the character being present in that place.To minimize errors, incorporating specific physical features of the specified race into the prompt is recommended. For instance, for a rabbit character, you can include rabbit ears and a rabbit tail. If you want only the ears to be animal, enter human ears in Negative positive. Since I struggle with hand drawings, I recommend using Embeddings or similar techniques. This model also capable of creating NSFW. I excel at crafting female characters but find males challenging. If you wish to generate only male characters, please input a 'negative prompt' example featuring pussy. I plan to fine-tune my skills to accurately depict male physical features in the future."Limitations of this model: The sense of distance with the character is buggy. Controlling it is extremely difficult. Please use the <a target="_blank" rel="ugc" href="https://civarchive.com/images/4620082">zoom slider!</a> (<a target="_blank" rel="ugc" href="https://civarchive.com/images/4620082">Zoom slider Sample 1</a>, <a target="_blank" rel="ugc" href="https://civarchive.com/images/4620081">Zoom slider Sample 2</a>)Be cautious as there is a potential for the character's body to turn blue. When specifying color prompts, please pay attention. Consider incorporating blue body, etc., in negative prompts.I wonder if someone can make a model that can generate a male pig anthropo 🐷<h3 id="heading-58">Rough Recipe:</h3>Merged using basic functions of simple merge and Super Merger and Lora merge. Combined based on intuition. In my merging process, I hardly consider the hierarchy; to put it simply, I adjust the concentration using both model and Lora for production. Apologies for not recalling specific details or values.<ul><li><a target="_blank" rel="ugc" href="https://note.com/plum_code/n/nef4062cacfe8#c12c998f-ee07-4be0-877f-1284758d4a0b">PlumBeer_FurryMix_v1</a></li><li><a target="_blank" rel="ugc" href="https://note.com/plum_code/n/n6dc166cb9920?magazine_key=m56c4c3dce14e#582f9575-bde9-4248-86e3-4e60d3b6a7d2">PlumBeer_FurryMix_v2</a></li><li><a target="_blank" rel="ugc" href="https://note.com/plum_code/n/n0579b45ca2eb#582f9575-bde9-4248-86e3-4e60d3b6a7d2">PlumBeer_FurryMIX_SuprCute_v1</a></li><li><a target="_blank" rel="ugc" href="https://note.com/plum_code/n/nd03f496d8224#607686ca-6187-479e-8a9b-105744deb966">PlumBeer_FurryMix_v3</a></li></ul>For those who want to merge with my model, I have a few tips. I started by merging <a target="_blank" rel="ugc" href="https://civarchive.com/models/34469?modelVersionId=163168">indigoFurryMix_v70AnimeNj5</a> and <a target="_blank" rel="ugc" href="https://civarchive.com/models/2583/hassaku-hentai-model">hassakuHentaiModel_v13</a>. Every time I merged Furry Lora, I added an appropriate amount and thoroughly wiped out human genes. Even if you merge with a model far from anthro, Fluffy won't disappear. It might be interesting to try merging with models other than Furry, and it's okay to take a little adventure. I admit I went a bit overboard with gene wiping, to the point of being cringe-worthy. But thanks to that, it became easier to merge based on v1, so it's all good in the end! Enjoy merging! And, of course, please respect the license!PS' Also, the bug in the distance perception with the character is embedded as a genetic factor. It can be reduced, but complete elimination is challenging. It's the result of my excessive emphasis on harmony between the landscape and characters. As the creator, I assure you. I'm also facing significant challenges.

PlumBeer FurryMIX

A more sophisticated version of v1

Files with SHA256: 193698c8ec1048bb5196f14b1bdc0e979b216f88f00f51ef209fee1d9e9bc601