<h3 id="ver-noob-v-pred-v3-eq"><strong>Ver </strong><span style="color:rgb(193, 194, 197)"><strong>NOOB V-PRED V3 EQ</strong></span></h3><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">Trained with improved parameters based on NOOB V-PRED EQ VAE exprimental version: </span><a target="_blank" rel="ugc" href="https://civarchive.com/models/1858821/noobai-v-pred-10-with-eq-vae">https://civarchive.com/models/1858821/noobai-v-pred-10-with-eq-vae</a></p><p></p><h3 id="updated-ver-noob-v-pred-v2">Updated Ver NOOB V-PRED V2</h3><p><span style="color:rgb(193, 194, 197)">Trained based on NOOB V-PRED 1.0 with OneTrainer. See about this version for training details.</span></p><p>It can do better wrinkles on pants and glans shape in this version. Sometimes it has text around but easily to remove with negative prompts.</p><p>The adaptability to art style has weakened in this version. In this case, upweight the style lora or doing a few more trials can be helpful.</p><p></p><h3 id="updated-ver-noob-v-pred-v1">Updated Ver NOOB V-PRED V1</h3><p>Trained based on NOOB V-PRED 1.0. Updated dataset to achieve better performance. It enhanced this concept on NOOB series models. See the examples first.</p><p>Only one trigger word in this version.</p><p></p><h3 id="updated">Updated</h3><p>Used PonyXL to train this one. Updated dataset to achieve better performance. It enhanced this concept on PonyXL and related models. See the examples first.</p><p><span style="color:rgb(193, 194, 197)">Seperated into two concepts: bulge down leg with pants/bulge down leg with dress.</span></p><p></p><h3 id="introduction">Introduction</h3><p>An experimental LyCORIS, used to generate futanari with bulge down leg.</p><p>Still a lot of room for improvement.</p><p></p><h3 id="pros">Pros</h3><ul><li><p>Can generate futanari with bulge down leg.</p></li></ul><h3 id="cons">Cons</h3><ul><li><p>Using 3D materials for training, the art style may become strange under high weights.</p></li><li><p>Penis twisted with legs.</p></li><li><p>Confused the glan with knees.</p></li><li><p>Bad penis shape.</p></li><li><p>A lot of trials needed.</p></li></ul><p></p><h3 id="how-to-use-it">How to use it</h3><p>I recommand using it in weight 0.7-0.8.</p><p></p><h3 id="training-details">Training details</h3><p>Traingset is of about 34 images. Half of them are chosen as regularization images. Total steps are about 1500.</p><p>resolution=768*768</p><p>batch_size=2</p><p>epoch=10</p><p>network_dim=16</p><p>network_alpha=16</p><p>conv_dim=4</p><p>conv_alpha=4</p><p>clip_skip=2</p><p>Using Prodigy as optimizer, setting as the author recommend.</p>

Files with SHA256: ddd4e951a162682f180ac794c3471e600b3e1b6cc9a690c207825763a176d766