<h2 id="heading-2084"><strong><span style="color:rgb(250, 82, 82)">Mariya</span><span style="color:rgb(238, 238, 238)"> also know as </span><span style="color:rgb(134, 142, 150)">Abbigail </span>/ <span style="color:rgb(134, 142, 150)">Svanhild</span></strong></h2><p><span style="color:rgb(238, 238, 238)">Russian glamour model born November 1986</span></p><h3 id="heading-344">Prompting:</h3><p>Replace token "<em>woman/girl/etc</em>" with trigger "<strong>mar1ya</strong>"</p><p><em>ex. Photo of mar1ya, doing something, wearing something, etc</em></p><h3 id="heading-309"><strong>Additional triggers</strong>:</h3><p>[long] blonde hair with bangs</p><h3 id="heading-310"><strong>Weight:</strong></h3><p>0.7 ~ 0.8</p><p>0.9 in Adetailer</p><h3 id="heading-311"><strong>Resolutions:</strong></h3><p><span style="color:rgb(193, 194, 197)">512x512 or 512x768 / 768x512</span></p><h3 id="heading-312">Clip Skip:</h3><p>1 for original model</p><h3 id="heading-345">Changelog</h3><p><strong>v1.1</strong></p><p>Dataset improved with new high-quality 2k resolution.</p><p>Opt-out from regularization images. </p><p>Clip skip change to 1 for more standardized use with realistic models.</p><p>No dynamic reduction of the DIM range used in this version... unfortunately.</p><p></p><p><strong>v1</strong></p><p>This Lora was not only trained to resemble her face but also her body. While I did the best, expect some novice mistakes, any suggestion will be well receive.</p><p>Regularization images used</p><h3 id="heading-2085"><em>Prompting:</em></h3><p>Photo of Mariya_Svanhild, "the rest of your prompts".</p><h3 id="heading-2086"><em>Weight:</em></h3><p>0.7 ~ 1</p><h3 id="heading-2087"><em>Specs:</em></h3><p>Images: 40<br />Repeats: 25<br />Regularization: 5000<br />Epoch: 2<br />Total train steps: 4000</p>

Files with SHA256: 52ed5cc5b19861efd0f7090417b230e82a1a1df3ae7d3d0e4dafd155dfb97a87