<p><span style="background-color:rgb(26, 27, 30);color:rgb(141, 141, 143);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">A LoRA specializing in flat color, no lineart, deformed character art.</span></p><p>Since this LoRA is made from a lossy differences from <a target="_blank" rel="ugc" href="https://civarchive.com/models/2846802/fwaun-tune">Fwaun Tune</a>, I recommend using these fine-tuned models unless there is a specific reason to use LoRA.<br />For details on each version, please refer to the release note of original fine-tuned model.</p><hr /><p><span style="background-color:rgb(26, 27, 30);color:rgb(141, 141, 143);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">フラットカラー・主線なし、（ちびというほどではない）デフォルメ画風を学習させたLoRAです。</span></p><p>このLoRAは<a target="_blank" rel="ugc" href="https://civarchive.com/models/2846802/fwaun-tune"> Fwaun Tune </a>のロッシーな差分から作られているので、特段の理由がなければファインチューンモデルのほうの利用を推奨します。</p><p>各バージョンの詳細は対応するオリジナルのファインチューンモデルのリリースノートをご確認ください。</p><h1 id="changelog">Changelog</h1><h2 id="krea2-v5.0">Krea2 v5.0</h2><p>モノクロと写実的な背景を出せるようになりました。<br />データセットやトレーニング時のパラメータについての詳細は<a target="_blank" rel="ugc" href="https://fwaunstp.fanbox.cc/posts/12400788">こちら（Fanbox）</a>。</p><p>You can now produce monochrome and realistic backgrounds.<br />For more details on the dataset and training parameters, <a target="_blank" rel="ugc" href="https://www.patreon.com/fwaunstp/posts/about-fwaun-tune-166332287?utm_medium=clipboard_copy&amp;utm_source=copyLink&amp;utm_campaign=postshare_creator&amp;utm_content=join_link">click here (Patreon)</a>.</p><h2 id="anima-v5.1"><strong>Anima v5.1</strong></h2><p>I thought it might be overfitting, so I lowered the learning rate.</p><p>過学習っぽい気がしたので学習率を下げました。</p><h2 id="anima-v5.0"><strong>Anima v5.0</strong></h2><p>You can now produce monochrome and realistic backgrounds.</p><p>モノクロと写実的な背景を出せるようになりました。</p><h1 id="about-anima-v4.0-krea2-v2.0-and-later">About Anima V4.0 / Krea2 v2.0 and Later</h1><p>Starting with this version, I have decided to create these LoRAs by extracting the differences between the fine-tuned model and the base model.</p><hr /><p>このバージョンから、このLoRAは対応するファインチューンモデルのベースモデルとの差分を抽出して作成するようになりました。</p><h1 id="before-v3.x">Before v3.x</h1><p>It uses the same dataset as <a target="_blank" rel="ugc" href="https://civarchive.com/models/2602206/fwaun-anima">Fwaun Anima</a> and <a target="_blank" rel="ugc" href="https://civarchive.com/models/2757203/fwaun-krea2">Fwaun Krea2</a>.</p><p>Unless there is a specific reason to use LORA, I recommend using <a target="_blank" rel="ugc" href="https://civarchive.com/models/2602206/fwaun-anima">Fwaun Anima</a> and <a target="_blank" rel="ugc" href="https://civarchive.com/models/2757203/fwaun-krea2">Fwaun Krea2</a>. instead of this LORA model.</p><h1 id="v3.1-krea2">v3.1 krea2</h1><ul><li><p>First lora for krea2.</p></li><li><p>Dataset images are not changed but captions are updated for training krea2.</p></li><li><p>Please place the trigger word at the end like comfyUI's official template.</p></li><li><p>Compared to the Anima model, the stylization is less deformed.</p></li><li><p>The "stylized proportion" tag falls between "realistic" and "chibi"</p></li></ul><h1 id="v3.0">v3.0</h1><p>To help body proportions, I’ve added super deformed, chibi-style illustrations to the training data.</p><p>You can decrease the body-to-head ratio using the “chibi, super deformed” tags and you can also use "realistic" tag to increase it.</p><p>Trigger word</p><pre><code>flat color, no lineart, deformed</code></pre><p>Optional tags</p><pre><code>chibi, super deformed</code></pre><pre><code>realistic</code></pre><h1 id="v2.0">v2.0</h1><p>To help adjust color tones and body proportions, I’ve added non-deformed flat color no lineart illustrations to the training data.</p><p>You can increase the body-to-head ratio using the “realistic” tag. If you don't want to increase it, set it to a negative value.</p><p>Trigger word</p><pre><code>flat color, no lineart, deformed</code></pre><p>Negative word</p><pre><code>chibi, realistic</code></pre><p></p>

Fwaun Style

<p>I thought it might be overfitting, so I lowered the learning rate.<br />For more details on the dataset and training parameters, <a target="_blank" rel="ugc" href="https://www.patreon.com/fwaunstp/posts/about-fwaun-tune-166332287?utm_medium=clipboard_copy&amp;utm_source=copyLink&amp;utm_campaign=postshare_creator&amp;utm_content=join_link">click here (Patreon)</a>.</p><p>過学習っぽい気がしたので学習率を下げました。<br />データセットやトレーニング時のパラメータについての詳細は<a target="_blank" rel="ugc" href="https://fwaunstp.fanbox.cc/posts/12400788">こちら（Fanbox）</a>。</p>

Files with SHA256: 240715117a263f93dfc00d8483bc38b3b3d2479af1c74ae4a4906cb924b317ec