<h2 id="about-the-lora">ℹAbout the LoRA</h2>This is a model of the Re:Zero characters + style.(aka Re:Zero − Starting Life in Another World, Re:Zero kara Hajimeru Isekai Seikatsu, Re: Life in a Different World from Zero)⚠Note: this is my first time making a character pack and it may have some issues.<hr /><span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);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">This model is<span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);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">:<ul><li>trained on 2D images (mostly on those that are close to the original style or from anime screenshots/official arts)</li><li>trained on some AI-generated images.</li><li>trained on SFW and NSFW data.</li><li>containing some data that was upscaled for better training</li></ul>Tips:<ul><li>you may want to lower the weight to make it a bit more flexible in some cases (0.7-0.9 should do the trick while keeping characters recognizable).</li><li>your checkpoint may already know some of the triggers below, so this LoRA will just increase the knowledge.</li></ul><hr /><h3 id="content-and-trigger-words">📦 Content and trigger words</h3>Some characters have less initial data and may work more poorly than others (sorted by high, medium, and low amounts of data below). I maybe will use the LoRA in the future to make more initial data for some of them (you can also help by sharing your clean anime-styled images of those characters).<ul><li>For style: rezeropackill<ul><li>you may also want to use anime screenshot.</li></ul></li><li>For characters:<ul><li>High amount of data:<ul><li>rem \(re:zero\)</li><li>ram \(re:zero\)</li><li>emilia \(re:zero\)</li><li>natsuki subaru</li><li>priscilla barielle</li><li>beatrice \(re:zero\)</li><li>elsa granhilte</li><li>crusch karsten</li><li>anastasia hoshin</li></ul>Medium amount of data:<ul><li>reinhard van astrea</li><li>julius juukulius</li><li>felt \(re:zero\)</li><li>felix argyle — as girl.</li><li>puck \(re:zero\) — as animal/cat.</li><li>wilhelm \(re:zero\) — also a small amount of data from his flashbacks (use old/old man or aged down)</li></ul></li><li>Low amount of data:<ul><li>petelgeuse romaneeconti</li><li>geuse \(re:zero\) — petelgeuse but before becoming insane.</li><li>roswaal l. mathers</li><li>aldebaran \(re:zero\)</li><li>rom \(re:zero\)</li><li>chin \(re:zero\)</li><li>kan \(re:zero\)</li><li>ton \(re:zero\)</li></ul></li></ul></li><li>(⚠Experimental: may not work) For outfits, you can use the same trigger word as the character but add outfit to the end (example: felt \(re:zero\) outfit). This can add an outfit for the original character when it's not already there, or you can try adding it to another character (which does not work that great). There is a few exceptions though:<ul><li>witch cult outfit</li><li>roswaal mansion maid uniform</li></ul></li></ul>💡 Note: trigger words don't require each other so you can use them separately (LoRA can currently sometimes work without the trigger words below, maybe I messed up somewhere? 🤔).Style also will be there along with the character, but you can add a style trigger word to make it stronger in some cases.<hr /><h3 id="if-you-like-this-model">🤝If you like this model</h3><ul><li>don't forget to give it a small like/tip</li><li>react to generated content</li><li>share your generations with others</li></ul>

Files with SHA256: 1a3f25d6303c91fb15b53c0517e37e7ec40c6e955affebb4fafa556a3fb7d89e