<h3 id="prompt-guide-()">✨ Prompt Guide (提示词指南)</h3>1. 🟢 Core Trigger (核心触发)<ul><li>Trigger Word: <code>sampo_koski</code></li><li>Expression (推荐表情): <code>smirk</code> (坏笑) 或 <code>winking</code> (眨眼)</li><li>Note: Facial features are baked in. 核心面部特征（蓝紫发、绿眼等）已深度绑定，无需额外输入发色/瞳色提示词。</li></ul>2. 🛡️ Default Costume (默认原皮服饰)<ul><li>Prompts: <code>sampo_koski, sampo_default_costume, 1boy, smirk, red_jacket, vest, pants, boots, belt, fingerless_gloves, shoulder_armor, asymmetrical_clothes, thigh_strap</code></li><li>Note: 极高精度还原游戏原画。请保留 <code>shoulder_armor</code> 和 <code>asymmetrical_clothes</code> 以稳定他标志性的不对称肩甲设计。</li></ul>3. 👕 Custom Outfit Mode (自由换装模式)<ul><li>How to use: Remove <code>sampo_default_costume</code> and gear tags to easily change clothes! 得益于严格的打标解耦，删掉 <code>sampo_default_costume</code> 及原皮细节词，即可完美换装。</li><li>Example (西装): <code>sampo_koski, 1boy, smirk, formal, black_suit, tie, holding_wine_glass</code></li></ul>4. 💣 Special Props (专属道具隔离)<ul><li>Note: 道具已在训练时做过特征隔离，可干净召唤，避免污染人物服饰。</li><li>Bomb (炸弹): <code>holding, bomb, heart_print</code></li><li>Money (钞票): <code>money, banknote, floating_money</code></li><li>Cards (卡牌): <code>playing_card, floating_cards</code></li></ul><hr /><h3 id="recommended-parameters-()">⚙️ Recommended Parameters (推荐生成参数)</h3><ul><li>LoRA Weight (权重): 0.8 - 1.0 (服饰已完全解耦，高权重不影响换装)</li><li>Resolution (分辨率): 1024 x 1024 / 832 x 1216</li><li>Sampler (采样器): Euler a / DPM++ 2M Karras</li><li>Steps (步数): 28 - 35</li><li>CFG Scale: 5 - 7</li><li>⚠️ Negative Prompts (核心负面词): <code>lowres, bad anatomy, bad hands, missing fingers, worst quality, (3d, 3d render, cgi, polygones:1.2)</code><ul><li>👉 Important: 训练集包含高质量多视角 3D 建模图。为了生成纯正的 2D 插画质感，请务必在负面词中加入 <code>3d, 3d render, cgi</code>！</li></ul></li></ul><hr /><h3 id="feedback-()">💬 Feedback (反馈)</h3>If you notice any issues (e.g., asymmetrical armor on the wrong side), please share your images in the comments! 如果你在出图时发现任何瑕疵（如不对称肩甲跑到了另一边、炸弹融合错误等），请务必带图在评论区告诉我。大家的返图将作为下个版本 (V2.0) 优化的最重要依据！Let's make it better together!

Files with SHA256: 51f8bf91c08df3ffab7e576a350a59d13cafc9b4729ee7bf06f48a727356184f