<h2 id="i-retrained-it-as-version-v2!-mxxezhq9q">I retrained it as version V2!</h2><h2 id="-a7medij57"></h2>Main Changes<ol><li>Improved clothing details, especially the accuracy of the mosaic-patterned outfit has increased.</li><li>Optimized prompts have been refined through multiple prompt recognition tests.</li></ol>This is a LoRA of the character "Anima" from Dungeon &amp; Fighter (DNF).I use it in ComfyUI by connecting it through nodes, so there are no specific trigger words.If you're using a tool other than ComfyUI, you should be able to load it using the <code>&lt;lora:~&gt;</code> syntax.It was trained on the ILXL base model, and I recommend setting the LoRA weight to around 0.8–1.0. For image generation, I used the WaiNSFW model as the checkpoint.<h3 id="base-prompt-2sgz20ofn">Base prompt</h3><pre><code> 1girl, white hair, yellow_eyes, orange pupils</code></pre><h3 id="outfit-prompt-7kfv4ce1u">Outfit prompt</h3><pre><code>black_bodysuit, bare_shoulder, white jacket, off-shoulder jacket</code></pre>If the bodysuit isn’t rendering well on the lower body, try adding black leggings!If bare feet aren’t showing properly, use barefoot.For the characteristic of clothing breaking apart in a mosaic pattern, use: square pattern, rectangle, disintegration clothing, digital_dissolved clothing, deconstructed clothing, fading outfit. '던전앤파이터'의 캐릭터 'Anima'의 Lora입니다. 저는 comfyUI에서 노드로 연결하여 Lora를 호출하여 사용하기에 트리거워드는 따로 없습니다. 만약 comfyUI가 아닌 다른 툴에서 사용 시에는 &lt;lora:~&gt; 하는 문법으로 호출하시면 사용 가능할 겁니다! 학습은 ILXL 베이스모델을 사용하였으며 추천하는 설정 값은 Lora weight 0.8~1.0 정도입니다. 그림을 뽑는 체크포인트의 경우 저의 경우엔 WaiNSFW 모델을 사용하였습니다. 베이스 프롬프트 <pre><code>1girl,white hair,yellow_eyes,orange pupils</code></pre>복장 프롬프트<pre><code>black_bodysuit,bare_shoulder,white jacket,off-shoulder jacket</code></pre>혹시 bodysuit가 하체쪽의 구현이 잘 되지 않는다면 black leggings 를 추가해주세요!맨발이 잘 나오지 않는다면 barefoot을 사용해주세요. 모자이크 패턴으로 분열되는 옷의 특징은 <pre><code>square pattern,rectangle,disintegration clothing,digital_dissolved clothing,deconstructed clothing,fading outfit</code></pre>를 사용하면 됩니다!

Files with SHA256: 8a7b3376c266291ba4642210e8e609f390297d265fc45294acd2e9857728a8da