<h1 id="style-collection">Style Collection</h1>This is my personal style collection. Anything could be here, except artist style.Cover images are directly from the vanilla (the original) base model in a1111, at default 1MP resolution. No upscale, no inpaint fixes. <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">Have metadata, 100% reproducible.All LoRAs in this collection, by default:<ul><li>Trained on illustrious v0.1.</li><li>Full natural language captions. No trigger words, just follow <s>your heart</s> cover images.</li><li>Version number means nothing, just my training job ID.</li></ul><hr />elpe v0.22:<ul><li>Dataset: Ultra-high-definition realistic renderings from game engine. Mainly portraits. 100% homemade. Full of lighting effects.</li><li>Style: Ultra realistic textures, light particles, cinematic lighting effects, fire effects, etc...<s>RTX: ON, quality: extreme, mode: path tracing</s> (lol, but also true)</li><li>Support both 2D anime and 3D game characters. Prefer 3D characters for best effect.</li><li>Potential trigger word: shattered lighting effect, vibrant lights, neon lights, glowing lines, light particles, fire... basically everything that emits light.</li><li>TBH, I did not believe SDXL can handle such complex dataset. But turns out it can, although not 100%, but looks cool enough.</li></ul>dcr v0.7:<ul><li>Dataset: Mixed dataset mainly from the game Azur Lane.</li><li>Style: Clean 2D anime illustrious, wallpaper.</li><li>Purpose: a style regulator for vanilla base model. Giving you the feeling of the game style (vibe, colors, clothing, etc.).</li></ul><hr />

Files with SHA256: 432ffad5d73e7b740b2b2574919da6b7257b38be7effeedc0641121d096d5891