<h2 id="cyrielle-sdxl-character-lora-(trigger:-cyriellegirl)">Cyrielle — SDXL Character LoRA (Trigger: CyrielleGirl)</h2>Trigger: <code>CyrielleGirl</code> Base: SDXL 1.0 (tested with <a rel="ugc" href="https://civarchive.com/models/312530/cyberrealistic-xl">CyberRealistic XL 7 </a>and <a rel="ugc" href="https://civarchive.com/models/277058?modelVersionId=1920523">epiCRealism XL VXVII-CrystalClear (Realism)</a>) Type: LoRACyrielle is my very first LoRA to produce a consistent female character LoRA (very pale skin, black or brunette hair, light gray eyes). The LoRA is designed to preserve face consistency while allowing diversity in outfits, poses, expressions, and contexts (office, outdoor, studio). It is built for realistic SDXL generations.<h2 id="recommended-settings">Recommended Settings</h2><ul><li>Sampler: DPM++ 2M Karras / Automatic</li><li>Steps: 32 (28–40 works well)</li><li>CFG: 5.5 (works between 4.5–6.5)</li><li>Resolution: 1024×1536</li><li>LoRA weight: 1.0 (use 0.7–0.9 if you want more influence from the base checkpoint)</li><li>ADetailer: enable for face (hands if needed)</li></ul><h2 id="prompt-examples">Prompt Examples</h2><ul><li>Clean portrait: <code>CyrielleGirl, ultra-detailed portrait, soft studio lighting, natural skin texture, shallow depth of field</code></li><li>Office look: <code>CyrielleGirl, elegant office outfit, seated at desk, laptop, morning window light, candid expression</code></li><li>Outdoor casual: <code>CyrielleGirl, casual outfit, denim jacket, walking on a tree-lined street, golden hour backlight</code></li><li>Full-body fashion: <code>CyrielleGirl, full-body studio shot, dynamic pose, softbox + rim light, editorial</code></li><li>Monochrome portrait: <code>CyrielleGirl, black and white portrait, dramatic Rembrandt lighting, high contrast, film grain</code></li><li>Negative prompt: <code>blurry, lowres, jpeg artifacts, extra fingers, bad hands, bad anatomy, watermark, text, childlike</code></li></ul><h2 id="training-notes">Training Notes</h2><ul><li>Dataset: portraits, half-body, and full-body shots in varied environments (350 img)</li><li>Goal: strong character consistency + adaptability across different contexts</li><li>Tools: SDXL, LoRA training, ADetailer cleanup for faces in sample renders</li></ul><h2 id="license-and-compliance">License &amp; Compliance</h2><ul><li>License: CC BY-NC 4.0 (non-commercial, credit appreciated).</li><li>No minors / underage / ambiguous age.</li><li>Mark NSFW content as such.</li></ul><h2 id="changelog">Changelog</h2><ul><li>v1.0: Initial public release</li></ul>

Files with SHA256: 89b8c2532362f8710304025778edbf408ddb599e91cf5a7c4f6cc0a9d387b87e