<h3>TLDR (Because I don't care)</h3><p>Use <code>Aglaya &lt;lora:Aglaya_v1:1&gt;</code> in positiv promt and <code>child</code> in nagative promt.</p><p><em>If you like the LoRa, please feel free to share your generated images here and leave a review (</em><strong><em>or at least a rating</em></strong><em>) ;D</em></p><p></p><h2>About</h2><p>Aglaya is one of these porn models that started in the early 201*, did a bunch of porn videos + photos and where never seen again...</p><p>I have a bunch of models on my list I wanted to put into a LoRa so I started with Aglaya first, she is cute and (at least on photos) has never seen nature ^^</p><p></p><h2>Training</h2><p>Well, I had selected 34 images from 5 photoshoots (that's all there is as far as I know) and cropped/photoshopped the guys out.</p><p>To be honest, I didn't think this would end well, simply because I thought it might lack variety... but it looks like I hit the sweet spot.</p><p>(I have to remember the ratio of nude to clothed images to partially train the breasts, but without overdoing it, which almost always resulted in nude generations).</p><p></p><h2>Usage</h2><p>No weight adjustments required so it is as simple as:</p><p><code>Aglaya &lt;lora:Aglaya_v1:1&gt;</code></p><p>In addition, I have made the best images using the <code>DPM++ SDE Karras</code> sampler with a fairly high number of steps - around <code>50 steps</code> is the sweet spot for me. Activating face fix depends on the situation, if the eyes are screwed up I always retry the same seed with face fix activated...</p><h2>Disclaimer</h2><p>All people in training and sample pictures are over the age of 18.</p>

Files with SHA256: 8f5195a2cc734c16eda2a46ac978bbfa34889897fa080c470a9e59b6c9fc4260