<h3 id="dcusa-or-style-lora">DCUSA | Style LoRA</h3>Issa new LoRA! Trained to create images in the style of DCUSA.<ul><li>Trigger Word: <code>dcusastyle</code></li><li>Recommended Weight: 0.8 - 1.0 Lower weights lean more subtle and soft, higher values emphasize facial style and body definition.</li><li>Additional Tags: <code>adult</code>, <code>bara</code>, <code>muscular</code></li></ul>Training Details<ul><li>Dataset of 225 manually curated, decensored and de-watermarked images</li><li>I did my best to remove as many speech bubbles and texts as possible (98% success!)</li><li>Tags were generated with a mix of automated and manual annotation</li><li>All sample images shown are generated in ComfyUI with a hires-fix pass (no inpainting or Adetailer)</li></ul>Usage Notes<ul><li>Works well with muscular and adult-themed prompts; add <code>bara</code> in your positive prompt if you want beefier results</li><li>To reduce unwanted generations, add <code>child, shota</code> to your negative prompt</li><li>NSFW material is part of the dataset. Safe outputs can be achieved by steering prompts away from NSFW content</li></ul>Please follow the <a target="_blank" rel="ugc" href="https://civarchive.com/safety">CivitAI Content Policy</a> when using this model.

Files with SHA256: 9a26b60db7e8aa8fd535a164eb4cc287fdf32cf00dd5cdd895e0c335e214e8cb