<h2 id="v2.0-ai-trained-(experimental)">🧪 v2.0 AI-Trained (Experimental)</h2>v2.0 was made with help from Claude Code (an AI assistant). It picked the training images, wrote the tags, ran the tests and chose the best epoch. It could be better or worse than v1.0, so try both and keep the one you like.<ul><li>Trigger word: <code>@sustyle</code> (note the @; v1.0 uses <code>sustyle</code>)</li><li>LoRA strength: 0.85 (0.7–0.75 for furry characters). Go up to 1.0 for a stronger style, but small pose details may not follow as closely.</li><li>Tested in: Forge Neo and ComfyUI (same results, about the same speed)</li><li>Settings: Euler a, Normal, 28 steps, CFG 4, Shift 3, 832×1216 or 1216×832</li><li>ComfyUI: euler_ancestral + normal. Shift 3 is the Anima default (no extra node). Load sustyle_e09 in a LoRA loader at 0.85 instead of the <code>&lt;lora:...&gt;</code> tag.</li></ul>Prompt layout (one idea per line, top to bottom):<pre><code>masterpiece, best quality, score_9, score_8, safe,
&lt;people count&gt;
&lt;character&gt;, &lt;series&gt;, @sustyle, &lt;hair, eyes, body&gt;,
&lt;pose&gt;,
&lt;outfit&gt;,
&lt;expression&gt;,
&lt;scene&gt;,
&lt;camera&gt; &lt;lora:sustyle_e09:0.85&gt;</code></pre>Change <code>safe</code> to <code>sensitive</code>, <code>nsfw</code> or <code>explicit</code> for spicier pictures.Negative prompt<pre><code>worst quality, low quality, lowres, score_1, score_2, score_3,
blurry, jpeg artifacts, watermark, signature, text, logo, (japanese text:1.2),
realistic, extra arms, extra legs, cropped feet, (emphasis lines:1.2), speech bubble</code></pre>Full guide with prompts, expressions, head angles and inpainting: <a target="_blank" rel="ugc" href="https://civarchive.com/articles/37183">Something Unlimited Style (Anima) — v2.0 Guide</a>ComfyUI workflow: a ready-made workflow (txt2img, inpaint, reuse prompt, hires fix, all with on/off toggles) is in the guide's Attachments: <a target="_blank" rel="ugc" href="https://civarchive.com/articles/37183">sustyle_v2_anima_workflow.json</a>.Pick v2.0 from the version tabs at the top to download it. The About this version box has the full settings.<hr /><h2 id="v1.0">v1.0</h2>Something Unlimited Anime Style LoRAThis is an anime-focused LoRA trained to capture the distinct, vibrant aesthetic of Something Unlimited. Whether you want to recreate that specific look or blend it into your existing anime workflows, this model provides a highly stylized, clean rendering effect.🔑 Trigger Word &amp; Usage<ul><li>Trigger Word: <code>sustyle</code></li><li>Flexibility: While <code>sustyle</code> is the dedicated trigger word, this LoRA is highly flexible. It can potentially work just as well without including the trigger word in your prompt, allowing you to subtly blend the style into your base model.</li></ul>⚠️ Important Prompting Tip (Fixing Lobo)Due to the dataset, prompting for <code>1boy</code> might occasionally cause the character Lobo to show up automatically.<ul><li>The Fix: If you do not want Lobo in your scene, add <code>(sulooboo:1.0 - 1.5)</code> to your negative prompt to clear him out.</li></ul>⚙️ Recommended Generation SettingsMost of the showcase images were generated using this exact configuration:<ul><li>Sampling Method: Euler a</li><li>Schedule Type: Normal</li><li>Sampling Steps: 28</li><li>CFG Scale: 4</li><li>Resolution: 832 x 1216 (Portrait)</li></ul><hr />🎨 Share Your Creations &amp; Feedback!I want to see how this performs across different prompts and base models!<ul><li>Post your generations: Try it out, generate some images, and upload your creations directly to this model page.</li><li>Leave feedback: Let me know what you think! Drop a review or comment with your thoughts so I can use your feedback for future updates.</li></ul><hr />

Something Unlimited Style - Anima

v2.0 AI-Trained (Experimental)This version was made with help from Claude Code (an AI assistant). It picked the training images, wrote the tags, ran the tests and chose the best epoch. That means it could be better or worse than v1.0, so try both and keep the one you like.Trigger word: <code>@sustyle</code> (v1.0 used <code>sustyle</code> without the @)Tested in: Forge Neo and ComfyUI (same results, about the same speed).Recommended settings<ul><li>Checkpoint: Anima (tested on dustVisionAnima v1.0)</li><li>LoRA strength: 0.85 (use 0.7–0.75 for furry characters)</li><li>Sampler: Euler a, Schedule: Normal</li><li>Steps: 28, CFG: 4, Shift: 3</li><li>ComfyUI: sampler euler_ancestral, scheduler normal. Shift 3 is the Anima default, so no extra node. Load sustyle_e09 in a LoRA loader at 0.85 instead of typing the lora tag.</li><li>Size: 832×1216 (portrait) or 1216×832 (landscape)</li></ul>Prompt layout (one idea per line, top to bottom):<pre><code>masterpiece, best quality, score_9, score_8, safe,
&lt;people count&gt;,
&lt;character&gt;, &lt;series&gt;, @sustyle, &lt;hair, eyes, body&gt;,
&lt;pose&gt;,
&lt;outfit&gt;,
&lt;expression&gt;,
&lt;scene&gt;,
&lt;camera&gt; &lt;lora:sustyle_e09:0.85&gt;</code></pre>Change <code>safe</code> to <code>sensitive</code>, <code>nsfw</code> or <code>explicit</code> for spicier pictures.Negative prompt<pre><code>worst quality, low quality, lowres, score_1, score_2, score_3,
blurry, jpeg artifacts, watermark, signature, text, logo, (japanese text:1.2),
realistic, extra arms, extra legs, cropped feet, (emphasis lines:1.2), speech bubble</code></pre>Quick tips<ul><li>Head turns: use <code>looking left</code> + <code>head turned left</code> (or right), and put the opposite side in the negative. <code>facing left/right</code> doesn't work.</li><li>Looks too strong or faces look off? Lower the strength to 0.7.</li><li>Fix small problems (hands, faces) with inpainting instead of rerolling the whole picture.</li></ul>ComfyUI workflow: a ready-made workflow (txt2img, inpaint, reuse prompt from an old picture, hires fix, all with on/off toggles) is in the guide's Attachments: sustyle_v2_anima_workflow.json.The full guide, with examples for expressions, head angles, inpainting and more, is in the article: <a target="_blank" rel="ugc" href="https://civitai.red/articles/37183">Something Unlimited Style (Anima) — v2.0 Guide</a>

Files with SHA256: e70da74bfa0a42fb3b98378558d6ba1ff1e51f952c6fd7bbf2c65cce8ba6a55b