<h1 id="heading-1455"><strong>Aesthetic Quality Modifiers - Absurdres</strong></h1><p></p><p><strong>(<span style="color:rgb(250, 176, 5)">Experimental</span>)</strong> This LoRA is part of an attempt to fine-tune a Pony Diffusion derived model to not require the use of score quality modifiers, with the hopes of also gaining more compatibility with other SDXL anime model merges.</p><p>Training data includes a selection of only top-scoring (&gt;75%) <code>absurdres</code> tagged booru images.</p><p>Manually (human) tagged in the same format of quality modifiers as the <a target="_blank" rel="ugc" href="https://civarchive.com/models/260267/animagine-xl-v31">Animagine XL V3.1 </a>checkpoint.</p><p>These modifiers can now be applied to the <strong>end of the prompt</strong></p><p></p><h3 id="heading-2234"><strong>Quality Modifiers</strong></h3><pre><code>
Quality Modifier	Score Criterion
masterpiece	        &gt; 95%
best quality	        &gt; 85% &amp; ≤ 95%
great quality	        &gt; 75% &amp; ≤ 85%
good quality	        &gt; 50% &amp; ≤ 75%
normal quality	        &gt; 25% &amp; ≤ 50%
low quality	        &gt; 10% &amp; ≤ 25%
worst quality	        ≤ 10%</code></pre><h3 id="heading-2237"><strong>Aesthetic Tags</strong></h3><pre><code>
Aesthetic Tag	       Score Range
very aesthetic	       &gt; 0.71
aesthetic	       &gt; 0.45 &amp; &lt; 0.71
displeasing	       &gt; 0.27 &amp; &lt; 0.45
very displeasing       ≤ 0.27</code></pre><h3 id="heading-175"><strong>Prompt structure</strong></h3><p>Positive prompt (no score tags required, quality modifiers at end of prompt):</p><pre><code>{{tags}}
masterpiece, best quality, very aesthetic, absurdres</code></pre><p>Negative prompt:</p><pre><code>low quality, worst quality, 3d, monochrome, (censor), render, text, watermark</code></pre>

Aesthetic Quality Modifiers - Absurdres

<p>Trained on <a target="_blank" rel="ugc" href="https://civitai.com/models/288584/autismmix-sdxl">AutismMix SDXL</a></p>

Files with SHA256: cb0bff30d1549c7d399266bb0305e9e63d483adb3ca4b1db2e31b7acb88b7fd3