<p><strong>NOTE: THE CHARACTER IMAGES ARE OF A CHILD 11-12 YEARS OLD.</strong></p><p><a target="_blank" rel="ugc" href="https://www.animecharactersdatabase.com/characters.php?id=89908">https://www.animecharactersdatabase.com/characters.php?id=89908</a></p><h2 id="heading-11">How to Use This Model</h2><p><strong>USE THEM SIMULTANEOUSLY</strong>. In this case, you need to download both <code>akashi_azurlane.pt</code> and <code>akashi_azurlane.safetensors</code>, then <strong>use </strong><code>akashi_azurlane.pt</code> as texture inversion embedding, and use <code>akashi_azurlane.safetensors</code> as LoRA at the same time.</p><p><strong>それらを同時に使用してください</strong>。この場合、<code>akashi_azurlane.pt</code>と<code>akashi_azurlane.safetensors</code>の両方をダウンロード する必要があります。<code>akashi_azurlane.pt</code>をテクスチャ反転埋め込みとして使用し、同時に<code>akashi_azurlane.safetensors</code>をLoRAとして使用してください。</p><p><strong>同时使用它们</strong>。在这种情况下，您需要下载<code>akashi_azurlane.pt</code>和<code>akashi_azurlane.safetensors</code>这两个文件，然后将<code>akashi_azurlane.pt</code>用作纹理反转嵌入， 同时使用<code>akashi_azurlane.safetensors</code>作为LoRA。</p><p>(Translated with ChatGPT)</p><p>The trigger word is <code>akashi_azurlane</code>, and the recommended tags are <code>masterpiece, best quality, highres, solo, {akashi_azurlane:1.10}, animal_ears, cat_ears, green_hair, long_hair, ahoge, yellow_eyes, hair_between_eyes, bow, very_long_hair, bell, hair_ornament, blush, bangs, :3, mole, hair_bow, mole_under_eye, smile</code>.</p><h2 id="heading-12">How This Model Is Trained</h2><p>This model is trained with <a target="_blank" rel="ugc" href="https://github.com/7eu7d7/HCP-Diffusion">HCP-Diffusion</a>. And the auto-training framework is maintained by <a target="_blank" rel="ugc" href="https://huggingface.co/deepghs">DeepGHS Team</a>.</p><h2 id="heading-13">Why Some Preview Images Not Look Like Akashi Azurlane</h2><p><strong>All the prompt texts</strong> used on the preview images (which can be viewed by clicking on the images) <strong>are automatically generated using clustering algorithms</strong> based on feature information extracted from the training dataset. The seed used during image generation is also randomly generated, and the images have not undergone any selection or modification. As a result, there is a possibility of the mentioned issues occurring.</p><p>In practice, based on our internal testing, most models that experience such issues perform better in actual usage than what is seen in the preview images. <strong>The only thing you may need to do is fine-tune the tags you use</strong>.</p><h2 id="heading-14">I Felt This Model May Be Overfitting or Underfitting, What Shall I Do</h2><p>Our model has been published on <a target="_blank" rel="ugc" href="https://huggingface.co/CyberHarem/akashi_azurlane">huggingface repository - CyberHarem/akashi_azurlane</a>, where models of all the steps are saved. Also, we published the training dataset on <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/CyberHarem/akashi_azurlane">huggingface dataset - CyberHarem/akashi_azurlane</a>, which may be helpful to you.</p><h2 id="heading-15">Why Not Just Using The Better-Selected Images</h2><p>Our model's entire process, from data crawling, training, to generating preview images and publishing, is <strong>100% automated without any human intervention</strong>. It's an interesting experiment conducted by our team, and for this purpose, we have developed a complete set of software infrastructure, including data filtering, automatic training, and automated publishing. Therefore, if possible, we would appreciate more feedback or suggestions as they are highly valuable to us.</p>

Files with SHA256: d022730ab82fabe0be9c4a5f0cfc8512cfb17dc43497133c0ab811bc07d2b445