<h1 id="trained-lora-for-akutox3's-art-style-on-sdxl.-built-with-67-curated-images-over-150-epochs.-2djxswhw2">Trained LoRA for Akutox3's art style on SDXL. Built with 67 curated images over 150 epochs.</h1><p>Works best at 1.0 strength, but can use 0.6 - 0.8 for subtle influence. Trigger: "Akutox3"</p><p>Training:</p><p>- 67 images × 150 epochs</p><p>- Dim/Conv: 32, Alpha: 16</p><p>- AdamW8bit + Huber loss</p><p>- Cosine w/restarts scheduler</p><p>- Full augmentations</p><p>- Base: noobAI Epsilon 1.0</p><p></p><h1 id="technical-observations-3pqulbuhr">Technical Observations</h1><h2 id="compatibility-and-style-b8epsfwua">Compatibility &amp; Style</h2><ul><li><p>Functions well with basic implementations without interference</p></li><li><p>Blends with other art styles, though consistency of style influence varies</p></li></ul><h2 id="known-limitations-yficjw4u4">Known Limitations</h2><ul><li><p>Weapon Rendering:</p><ul><li><p>Good performance with swords</p></li><li><p>Poor knife rendering - note that this may be due to common limitations across many art styles rather than specific to Akutox LoRA</p></li><li><p>Results may vary; your experience could differ</p></li></ul></li><li><p>Pose Constraints:</p><ul><li><p>Tendency to default to sitting positions</p></li><li><p>Particularly noticeable with prompts involving animals on arms</p></li><li><p>May be related to either LoRA or base model behavior (unconfirmed)</p></li><li><p>Individual results may vary depending on specific prompts and settings</p></li></ul></li></ul><h2 id="integration-notes-th9trxujv">Integration Notes</h2><ul><li><p>Successfully maintains basic functionality</p></li><li><p>Adaptable with other styles while preserving core characteristics</p></li><li><p>Note: All observations are based on specific testing conditions and may not represent all possible use cases</p></li></ul><p>Enjoy!</p>

Files with SHA256: b3e5ad19eecdb62a0a2d0a4c326d4af1333e402ad724c4f0fad43618050c8388