RC-AIO 2.9B / RealCosplay All In One Anima2.9B<pre><code>All-in-One Multi-Style Base Model (Anima 2.9B Full Fine-tune)A full fine-tune of Anima 2.9B designed as a true multi-style (Realistic / 3D / Anime) All-in-One base model.It keeps the strong anime foundation of Anima by default, while high-quality real cosplay and scenery photos are used to improve anatomy, composition, and lighting. Realistic capability is significantly enhanced without destroying the original anime concept.</code></pre>Improvements over 2.0B<pre><code>Realistic style fits noticeably better, and limbs/fingers are more stable and natural. This is currently my most satisfying version.</code></pre>Recommended SettingsPositive (Base)<pre><code>masterpiece, best quality, highres,beauty face</code></pre>Positive for Realistic<pre><code>photo \(medium\) 【0.5-1.5】</code></pre>Negative<pre><code>worst quality, low quality, lowres, jpeg artifacts, mature female, fat, aegyo sal, puffy cheeks, ugly face, thick lips, </code></pre>Parameters<ul><li>Sampler: ER SDE / Euler</li><li>Schedule: beta</li><li>CFG: 3~4 (I recommend 3.5)</li><li>Shift: 3</li></ul>Hires.fix / Upscale Tip If using secondary sampling or Hires.fix, keep the upscale multiplier at 1.8x or lower. Higher values can improve finger details but tend to introduce more noise.Artist Style @danbooru artist nameSpecially Trained Tags<ul><li>Background: blurry foreground, blurry background…</li><li>Art Style: photo \(medium\), figure \(medium\), realistic, 3D, impasto, draft, flat color, no lineart, @nyalia, @eknn…</li><li>Lighting: dark, chiaroscuro, night, backlighting, overexposure… (pls use "chiaroscuro" instead of "camera flash")</li><li>Composition: fisheye, partially underwater shot, dutch angle, pov crotch, foreshortening…</li><li>Age: … , man/woman, mature male/female, old man/woman, …</li><li>Others: aegyo sal, puffy cheeks, bloom, vignetting, chromatic aberration, etc.</li></ul>Not Recommended Tag<ul><li>solo (removed from training data, use solo focus instead)</li><li>Duplicate synonymous tags (e.g. <s>shirt</s>, red shirt).These were filtered out during training to improve prompt adherence.</li></ul>

Files with SHA256: 87c8293ed9667ae9b348db283a95a0f718c125ea62d8dfae37c9190c513098ca