<h3 id="heading-2299">I'm back with another dumb loRA :D</h3>This time it's a style loRA <s>(aka an attempt of mine)</s>Most images in the dataset come from <a target="_blank" rel="ugc" href="https://hub.vroid.com/tags/VRoidStudio/artworks">VRoid Hub</a> while others were made by me in VRoid Studio so yeah<h1 id="heading-2300">SOME DUMB INFO:</h1><ul><li>trained on 87 handpicked images (some were homemade)</li><li>works best with <a target="_blank" rel="ugc" href="https://civarchive.com/models/23900/anylora-checkpoint">AnyLoRA</a></li><li>when doing full bodies or cowboy shots, use a higher res then 512x512 or turn on hires. fix</li><li>tags such as "Cel Shading" or "Anime" might help</li><li>I haven't tested this with NSFW stuff so I'm not sure if this loRA can even do NSFW content <s>(probably not since all dataset images are sfw)</s>. Feel free to play around with the model tho.</li><li>Feel free to play with the loRA weight. 1 should be enough tho</li><li>to do male characters, you might need to force it (example: instead of "1boy", use "(1boy:1.4)")</li></ul><h1 id="heading-2301">CONTACT INFO!:</h1>Twitter: <a target="_blank" rel="ugc" href="https://twitter.com/YuriMayori">YuriMayori</a>DeviantArt: <a target="_blank" rel="ugc" href="https://www.deviantart.com/cappyadams">CappyAdams</a>Other Socials (Link to my Website): <a target="_blank" rel="ugc" href="https://cappyadams.github.io/index.html">Click here or smthin idk</a>

VRoid Studio Model Style // LoRA

"3d, anime coloring" tags seem to get a more accurate style to a traditional Vroid model. experimental

Files with SHA256: e60ee4ec4b0fe68be1b470dc79125c22573d73e36ad3a1e0b1f36471db2e821a