<h3 id="z-image-base-turbo-int8">Z-Image Base / Turbo INT8 优化版</h3>这是我制作的 Z-Image Base 和 Z-Image Turbo 混合精度 INT8 量化版本，可直接在新版 ComfyUI 中使用。量化时保留了敏感模块及部分关键首尾层，以减少毛边、噪点和细节损失，同时降低显存占用并提高生成速度。在 <code>1024 × 1024</code> 分辨率下个人实测：<ul><li>BF16：约 <code>1.66s/步</code></li><li>普通 INT8（不额外保留层）：约 <code>0.85s/步</code></li><li>混合 INT8（保留关键层）：约 <code>1.00s/步</code></li></ul>混合版本比普通 INT8 稍慢，但画质更接近 BF16，是速度和质量之间更均衡的选择。实际性能会因显卡和工作流而异。 Z-Image Base / Turbo INT8 OptimizedThese are mixed-precision INT8 quantized versions of Z-Image Base and Z-Image Turbo, compatible with the latest ComfyUI.Sensitive modules and several key early/late layers are kept at higher precision to reduce jagged edges, noise, and detail loss while improving speed and reducing VRAM usage.Personal benchmark at <code>1024 × 1024</code>:<ul><li>BF16: <code>1.66s/step</code></li><li>Standard INT8 without extra preserved layers: <code>0.85s/step</code></li><li>Mixed INT8 with key layers preserved: <code>1.00s/step</code></li></ul>The mixed INT8 version is slightly slower than standard INT8, but provides quality closer to BF16, offering a better balance between speed and image quality. Performance may vary depending on your GPU and workflow. A little support goes a long way! If you’d like to help me keep creating, you can do so at <a target="_blank" rel="ugc" href="https://ko-fi.com/xabsurd">https://ko-fi.com/xabsurd</a><ul><li>获得更多未公开内容，和提前得至少1个月到更高质量lora输出</li><li>Get more undisclosed content, and get at least 1 month ahead of time to higher quality lora output</li></ul><a target="_blank" rel="ugc" href="https://ko-fi.com/xabsurd"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/38b3106b-e093-40aa-89e7-42183d6f22bd/original=true/38b3106b-e093-40aa-89e7-42183d6f22bd.jpeg" /></a>

Files with SHA256: f334923c84c93be395a4eff0c33ed4d15c8b73f56f1655d1f154cf03647f1742