<h1>The goal of this repo is to</h1> <ul> <li>Capturing webtoon character's unique characteristics</li> <li>Get varieties of poses, gestures and actions without damaging too many characteristics</li> </ul>

<h3>For the LoRA inference</h3> <ul> <li>Current LoRA checkpoints are under development. Instruction will be added soon</li> <li>For those who want to try out. I recommend <b>Midnight Mixers</b> as the base model.</li> <li><b>512(width) x 640(height)</b></li> <li><b>50 steps / 7 cfg</b>. Step below 40 would yield poor quality</li> <li><b>0.5~0.6 range LoRA weight.</b></li> </ul>