<h1 id="v3.0(utimate)">V3.0(Utimate)</h1><h3 id="my-pixiv:-ghostshadow-pixiv">My Pixiv: <a target="_blank" rel="ugc" href="https://www.pixiv.net/users/62701726">GhostShadow - pixiv</a></h3><h3 id="the-previous-update-didn't-satisfy-me.-after-retraining-and-optimizing-the-process-i've-released-this-refined-version-which-best-captures-the-art-style-of-the-milk-factory-game-cg-style.-if-you-like-milk-factory-game-cg-style-but-prefer-a-milder-version-the-first-two-releases-are-also-good-options.">The previous update didn't satisfy me. After retraining and optimizing the process, I've released this refined version, which best captures the art style of the Milk factory Game CG Style. If you like Milk factory Game CG Style but prefer a milder version, the first two releases are also good options.</h3>This LoRA is based on the game CGs from four works by the Milk Factory. When it comes to processing the dataset, I pulled out the main image from each CG—（excluding any pictures where, like, semen covers the whole body, or those with extra frames）—and carefully selected 413 images for training.I noticed that most of the existing Milk Factory models are based on SD1.5 or Pony, so I trained this LoRA on top of Illustrious, hoping it'll give you the results you're looking for.<h3 id="recommended-settings:">Recommended settings:</h3><ul><li>20 to 30 steps for the number of steps</li><li>CFG scale of 7.0</li><li>using the Euler A sampling method.</li><li>You can mix it with other styles too, which might give you even better results.</li></ul>===============================================这个Lora是基于Milk factory的四部作品的游戏CG而制成。对于数据集的处理，我将每张CG中的最主要的一张（将比如精液覆盖全身的图片去除，或者有额外画面的图片去除）提取出来，精心选出了413张图片进行训练。我发现现有的milk factory大多都是基于SD1.5或者pony，因而我训练了基于illustrious的Lora，希望你可以通过此达到你想要的效果<h2>推荐设置：</h2><ul><li>步数20-30步</li><li>CFG：7.0</li><li>采样方法：Euler A</li><li>可以和其他style混合使用，可以达到更好的效果</li></ul>

Milk factory Game CG Style (もっと！孕ませ！)

I will update along with the progress of the game.我将会随着游戏进度一同更新

Files with SHA256: 05cae30eed25abaa980107812c1d6be327ee1eb1265695812bd3e0e979c8366e