This style model is based on the style of <a target="_blank" rel="ugc" href="https://x.com/zzzxxxccc_18">zzzxxxccc</a>, a futanari-focused 3D artist. If the original artist requests a takedown, I will remove this model from CivitAI. Please support <a target="_blank" rel="ugc" href="https://x.com/zzzxxxccc_18">zzzxxxccc</a> at their Patreon <a target="_blank" rel="ugc" href="http://patreon.com/threezxc">here</a>! Please do not monetize your generations. Enjoy, and feel free to post generations you’re happy with. Thank you!<hr />All sample images were generated locally using the recommended settings listed below, plus face/hand/eye detailer steps and an upscale using Mixture of Diffusers in Forge Neo.<h3 id="settings-and-recommendations">Settings &amp; Recommendations</h3><ul><li>Trigger Word: @3zxc, realistic,<ul><li>Try adding "nose, lips," if anime/cartoon characters' faces still look too 2D.</li></ul></li></ul><ul><li>Strength: 1.0</li><li>Checkpoint: <a target="_blank" rel="ugc" href="https://civarchive.com/models/2458426/anima?modelVersionId=2945208">Anima Base-v1.0</a></li><li>Use with <a target="_blank" rel="ugc" href="https://civarchive.com/models/2364703/rdbt-anima?modelVersionId=3017013">RDBT b1 v0.39.b</a> to get the results from the example images! Anima struggles a bit with 3D/realistic styles out of the box. Run RDBT @ strength 1.0 alongside this style or you will get different results, but experiment to see what you prefer. Alternatively, if you use the <a target="_blank" rel="ugc" href="https://civarchive.com/models/2356447/rdbt-or-anima?modelVersionId=3016541">RDBT checkpoint</a> instead of vanilla Anima Base, you won't have to use the additional RDBT LoRA (as it's baked into that checkpoint). Don't stack the RDBT LoRA with the RDBT checkpoint! See the different settings/tags needed below with and without RDBT.</li><li>Reduce character tag strength! Some characters have a particular style baked into their tag. Experiment with reducing the strength of character tags to somewhere between 0.5 &amp; 1.0 to get the character without much of the style Anima has baked in for them. Basically, instead of "Shantae," use something like "(Shantae:0.5),".</li></ul><ul><li>Sampler: ER SDE</li><li>Scheduler: Simple</li><li>With <a target="_blank" rel="ugc" href="https://civarchive.com/models/2364703/rdbt-anima?modelVersionId=3017013">RDBT</a>:<ul><li>Positive: masterpiece, @3zxc, realistic,</li><li>Negative: worst quality, low quality, bad anatomy, lowres, anime coloring, low poly</li><li>CFG: 3 or less</li><li>Steps: 24</li></ul></li><li>Without <a target="_blank" rel="ugc" href="https://civarchive.com/models/2364703/rdbt-anima?modelVersionId=3017013">RDBT</a>:<ul><li>Positive: masterpiece, best quality, score_7, @3zxc, realistic,</li><li>Negative: worst quality, low quality, score_1, score_2, score_3, bad anatomy, lowres, anime coloring, low poly</li><li>CFG: 4</li><li>Steps: 30</li></ul></li></ul><ul><li>HiResFix/MultiDiffusion (optional):<ul><li>Upscaling Model: <a target="_blank" rel="ugc" href="https://openmodeldb.info/models/1x-ITF-SkinDiffDDS-v1">1x_ITF_SkinDiffDDS_v1.pth</a></li><li>Upscale By: 1.50x - 1.75x (anything above 1.75 starts to break down)</li><li>Denoising Strength: 0.03 - 0.05 (anything much higher than this may just make things look worse, but feel free to try)</li></ul></li><li>Adetailer (optional):<ul><li>Denoising Strength: For face, eyes, &amp; hands, recommend about 0.10. For genitals, recommend 0.03.</li><li>In Forge Neo, set "Inpaint only masked" to at least 64px (I just max it out to 256px). This gives the model the ability to see more context around the thing being detailed. A higher number here means a better result (more accurate color, less jank).</li><li>If you have trouble with the eye detailer step (especially for an image where both eyes aren't clearly visible), try individual eye detection instead of paired. Individual and paired versions of eye detection models are in the link below. </li><li>Face segmentation <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/Anzhcs_YOLOs/blob/main/Anzhc%20Face%20seg%201024%20v2%20y8n.pt">model link</a></li><li>Eye detection <a target="_blank" rel="ugc" href="https://civarchive.com/models/178518/eyeful-or-robust-eye-detection-for-adetailer-comfyui?modelVersionId=582139">model link</a></li><li>Hand segmentation <a target="_blank" rel="ugc" href="https://civarchive.com/models/329458/hand-detailersegmentation-adetailer?modelVersionId=1780243">model link</a></li><li>Genitalia detection <a target="_blank" rel="ugc" href="https://civarchive.com/models/132388/penis-pussy-adetailer-model?modelVersionId=149639">model link</a></li></ul></li></ul>

zzzxxxccc Style | Anima

Use with <a target="_blank" rel="ugc" href="https://civitai.red/models/2364703/rdbt-anima?modelVersionId=3017013">RDBT b1 v0.39.b</a> to get the same results as the example images. See the description for recommended settings when using this style with &amp; without RDBT. This GLoRA was trained on <a target="_blank" rel="ugc" href="https://civitai.red/models/2458426/anima?modelVersionId=2945208">Anima Base-v1.0</a>."v4" refers to version 4 of the dataset, which has been updated since the NoobAI <a target="_blank" rel="ugc" href="https://civitai.red/models/2131692/zzzxxxccc-style-or-ilnoobai?modelVersionId=2920481">v3-EPS</a> version was trained.See the Optional Files section for the training settings .toml file.Several different training settings were attempted until this version finally managed to capture the style without completely falling apart. Anima is first and foremost for non-photorealistic styles, so getting a detailed, semirealistic/3D style to work reliably ended up requiring making this GLoRA a larger filesize than was initially expected.

Files with SHA256: 1523543a47db59675496b041b9744bc5cc0b446eb8e1dc0b759be7d07098e711