diffusers/pr_13966/zh/hybrid_inference/vae_encode.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;入门：使用混合推理进行 VAE 编码&quot;,&quot;local&quot;:&quot;入门使用混合推理进行-vae-编码&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;内存&quot;,&quot;local&quot;:&quot;内存&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;可用 VAE&quot;,&quot;local&quot;:&quot;可用-vae&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;代码&quot;,&quot;local&quot;:&quot;代码&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;基本示例&quot;,&quot;local&quot;:&quot;基本示例&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;生成&quot;,&quot;local&quot;:&quot;生成&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;集成&quot;,&quot;local&quot;:&quot;集成&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
		<link href="/docs/diffusers/main/zh/_app/immutable/entry/start.BjXnKLdN.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/BqKw7OfG.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/BbekZcyp.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/entry/app.C861NjaH.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/vgYqWUA9.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/JwskCAIl.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/DspwCNYP.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/nodes/0.DGQv4KCz.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/D9q6WOrq.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/nodes/9.jEhcNG5Q.js" rel="modulepreload">
		<link href="/docs/diffusers/main/zh/_app/immutable/chunks/DdZvggmf.js" rel="modulepreload">
		<!--1wq4egn--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;入门：使用混合推理进行 VAE 编码&quot;,&quot;local&quot;:&quot;入门使用混合推理进行-vae-编码&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;内存&quot;,&quot;local&quot;:&quot;内存&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;可用 VAE&quot;,&quot;local&quot;:&quot;可用-vae&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;代码&quot;,&quot;local&quot;:&quot;代码&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;基本示例&quot;,&quot;local&quot;:&quot;基本示例&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;生成&quot;,&quot;local&quot;:&quot;生成&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;集成&quot;,&quot;local&quot;:&quot;集成&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/><!---->
		<link href="/docs/diffusers/main/zh/_app/immutable/assets/0.tn0RQdqM.css" rel="modulepreload"> <!--[--><!--[0--><!--[--><!--[0--><!--[--><p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"><div class="inline-flex rounded-md max-sm:rounded-sm"><button class="inline-flex items-center gap-1 h-7 max-sm:h-7 px-2 max-sm:px-1.5 text-sm font-medium text-gray-800 border border-r-0 rounded-l-md max-sm:rounded-l-sm border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-live="polite"><span class="inline-flex items-center justify-center rounded-md p-0.5 max-sm:p-0 hover:text-gray-800 dark:hover:text-gray-200"><svg class="sm:size-3.5 size-3" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----></span> <span>Copy page</span></button> <button class="inline-flex items-center justify-center w-6 max-sm:w-5 h-7 max-sm:h-7 disabled:pointer-events-none text-sm text-gray-500 hover:text-gray-700 dark:hover:text-white rounded-r-md max-sm:rounded-r-sm border border-l transition border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-haspopup="menu" aria-expanded="false" aria-label="Open copy menu"><svg class="transition-transform text-gray-400 overflow-visible sm:size-3.5 size-3 rotate-0" width="1em" height="1em" viewBox="0 0 12 7" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M1 1L6 6L11 1" stroke="currentColor"></path></svg><!----></button></div> <!--[-1--><!--]--></div><!----> <!--[0--><h1 class="relative group"><a id="入门使用混合推理进行-vae-编码" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#入门使用混合推理进行-vae-编码"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>入门：使用混合推理进行 VAE 编码</span></h1><!--]--><!----> <p>VAE 编码用于训练、图像到图像和图像到视频——将图像或视频转换为潜在表示。</p> <!--[1--><h2 class="relative group"><a id="内存" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#内存"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>内存</span></h2><!--]--><!----> <p>这些表格展示了在不同 GPU 上使用 SD v1 和 SD XL 进行 VAE 编码的 VRAM 需求。</p> <p>对于这些 GPU 中的大多数，内存使用百分比决定了其他模型（文本编码器、UNet/Transformer）必须被卸载，或者必须使用分块编码，这会增加时间并影响质量。</p> <details><summary>SD v1.5</summary> <table><thead><tr><th align="left">GPU</th><th align="left">分辨率</th><th align="right">时间（秒）</th><th align="right">内存（%）</th><th align="right">分块时间（秒）</th><th align="right">分块内存（%）</th></tr></thead><tbody><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">512x512</td><td align="right">0.015</td><td align="right">3.51901</td><td align="right">0.015</td><td align="right">3.51901</td></tr><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">256x256</td><td align="right">0.004</td><td align="right">1.3154</td><td align="right">0.005</td><td align="right">1.3154</td></tr><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">2048x2048</td><td align="right">0.402</td><td align="right">47.1852</td><td align="right">0.496</td><td align="right">3.51901</td></tr><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">1024x1024</td><td align="right">0.078</td><td align="right">12.2658</td><td align="right">0.094</td><td align="right">3.51901</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">512x512</td><td align="right">0.023</td><td align="right">5.30105</td><td align="right">0.023</td><td align="right">5.30105</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">256x256</td><td align="right">0.006</td><td align="right">1.98152</td><td align="right">0.006</td><td align="right">1.98152</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">2048x2048</td><td align="right">0.574</td><td align="right">71.08</td><td align="right">0.656</td><td align="right">5.30105</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">1024x1024</td><td align="right">0.111</td><td align="right">18.4772</td><td align="right">0.14</td><td align="right">5.30105</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">512x512</td><td align="right">0.032</td><td align="right">3.52782</td><td align="right">0.032</td><td align="right">3.52782</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">256x256</td><td align="right">0.01</td><td align="right">1.31869</td><td align="right">0.009</td><td align="right">1.31869</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">2048x2048</td><td align="right">0.742</td><td align="right">47.3033</td><td align="right">0.954</td><td align="right">3.52782</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">1024x1024</td><td align="right">0.136</td><td align="right">12.2965</td><td align="right">0.207</td><td align="right">3.52782</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">512x512</td><td align="right">0.036</td><td align="right">8.51761</td><td align="right">0.036</td><td align="right">8.51761</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">256x256</td><td align="right">0.01</td><td align="right">3.18387</td><td align="right">0.01</td><td align="right">3.18387</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">2048x2048</td><td align="right">0.863</td><td align="right">86.7424</td><td align="right">1.191</td><td align="right">8.51761</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">1024x1024</td><td align="right">0.157</td><td align="right">29.6888</td><td align="right">0.227</td><td align="right">8.51761</td></tr><tr><td align="left">NVIDIA GeForce RTX 3070</td><td align="left">512x512</td><td align="right">0.051</td><td align="right">10.6941</td><td align="right">0.051</td><td align="right">10.6941</td></tr><tr><td align="left">NVIDIA GeForce RTX 3070</td><td align="left">256x256</td><td align="right">0.015</td><td align="right"></td><td align="right"></td><td align="right"></td></tr><tr><td align="left">3.99743</td><td align="left">0.015</td><td align="right">3.99743</td><td align="right"></td><td align="right"></td><td align="right"></td></tr><tr><td align="left">NVIDIA GeForce RTX 3070</td><td align="left">2048x2048</td><td align="right">1.217</td><td align="right">96.054</td><td align="right">1.482</td><td align="right">10.6941</td></tr><tr><td align="left">NVIDIA GeForce RTX 3070</td><td align="left">1024x1024</td><td align="right">0.223</td><td align="right">37.2751</td><td align="right">0.327</td><td align="right">10.6941</td></tr></tbody></table></details> <details><summary>SDXL</summary> <table><thead><tr><th align="left">GPU</th><th align="left">Resolution</th><th align="right">Time (seconds)</th><th align="right">Memory Consumed (%)</th><th align="right">Tiled Time (seconds)</th><th align="right">Tiled Memory (%)</th></tr></thead><tbody><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">512x512</td><td align="right">0.029</td><td align="right">4.95707</td><td align="right">0.029</td><td align="right">4.95707</td></tr><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">256x256</td><td align="right">0.007</td><td align="right">2.29666</td><td align="right">0.007</td><td align="right">2.29666</td></tr><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">2048x2048</td><td align="right">0.873</td><td align="right">66.3452</td><td align="right">0.863</td><td align="right">15.5649</td></tr><tr><td align="left">NVIDIA GeForce RTX 4090</td><td align="left">1024x1024</td><td align="right">0.142</td><td align="right">15.5479</td><td align="right">0.143</td><td align="right">15.5479</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">512x512</td><td align="right">0.044</td><td align="right">7.46735</td><td align="right">0.044</td><td align="right">7.46735</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">256x256</td><td align="right">0.01</td><td align="right">3.4597</td><td align="right">0.01</td><td align="right">3.4597</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">2048x2048</td><td align="right">1.317</td><td align="right">87.1615</td><td align="right">1.291</td><td align="right">23.447</td></tr><tr><td align="left">NVIDIA GeForce RTX 4080 SUPER</td><td align="left">1024x1024</td><td align="right">0.213</td><td align="right">23.4215</td><td align="right">0.214</td><td align="right">23.4215</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">512x512</td><td align="right">0.058</td><td align="right">5.65638</td><td align="right">0.058</td><td align="right">5.65638</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">256x256</td><td align="right">0.016</td><td align="right">2.45081</td><td align="right">0.016</td><td align="right">2.45081</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">2048x2048</td><td align="right">1.755</td><td align="right">77.8239</td><td align="right">1.614</td><td align="right">18.4193</td></tr><tr><td align="left">NVIDIA GeForce RTX 3090</td><td align="left">1024x1024</td><td align="right">0.265</td><td align="right">18.4023</td><td align="right">0.265</td><td align="right">18.4023</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">512x512</td><td align="right">0.064</td><td align="right">13.6568</td><td align="right">0.064</td><td align="right">13.6568</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">256x256</td><td align="right">0.018</td><td align="right">5.91728</td><td align="right">0.018</td><td align="right">5.91728</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">2048x2048</td><td align="right">内存不足 (OOM)</td><td align="right">内存不足 (OOM)</td><td align="right">1.866</td><td align="right">44.4717</td></tr><tr><td align="left">NVIDIA GeForce RTX 3080</td><td align="left">1024x1024</td><td align="right">0.302</td><td align="right">44.4308</td><td align="right">0.302</td><td align="right">44.4308</td></tr><tr><td align="left">NVIDIA GeForce RTX 3070</td><td align="left">512x512</td><td align="right">0.093</td><td align="right">17.1465</td><td align="right">0.093</td><td align="right">17.1465</td></tr></tbody></table> <p>| NVIDIA GeForce R
| NVIDIA GeForce RTX 3070       | 256x256      |            0.025 |               7.42931 |                  0.026 |            7.42931 |
| NVIDIA GeForce RTX 3070       | 2048x2048    |          OOM     |             OOM       |                  2.674 |           55.8355  |
| NVIDIA GeForce RTX 3070       | 1024x1024    |            0.443 |              55.7841  |                  0.443 |           55.7841  |</p></details> <!--[1--><h2 class="relative group"><a id="可用-vae" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#可用-vae"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>可用 VAE</span></h2><!--]--><!----> <table><thead><tr><th align="center"></th><th align="center"><strong>端点</strong></th><th align="center"><strong>模型</strong></th></tr></thead><tbody><tr><td align="center"><strong>Stable Diffusion v1</strong></td><td align="center"><a href="https://qc6479g0aac6qwy9.us-east-1.aws.endpoints.huggingface.cloud" rel="nofollow">https://qc6479g0aac6qwy9.us-east-1.aws.endpoints.huggingface.cloud</a></td><td align="center"><a href="https://hf.co/stabilityai/sd-vae-ft-mse" rel="nofollow"><code>stabilityai/sd-vae-ft-mse</code></a></td></tr><tr><td align="center"><strong>Stable Diffusion XL</strong></td><td align="center"><a href="https://xjqqhmyn62rog84g.us-east-1.aws.endpoints.huggingface.cloud" rel="nofollow">https://xjqqhmyn62rog84g.us-east-1.aws.endpoints.huggingface.cloud</a></td><td align="center"><a href="https://hf.co/madebyollin/sdxl-vae-fp16-fix" rel="nofollow"><code>madebyollin/sdxl-vae-fp16-fix</code></a></td></tr><tr><td align="center"><strong>Flux</strong></td><td align="center"><a href="https://ptccx55jz97f9zgo.us-east-1.aws.endpoints.huggingface.cloud" rel="nofollow">https://ptccx55jz97f9zgo.us-east-1.aws.endpoints.huggingface.cloud</a></td><td align="center"><a href="https://hf.co/black-forest-labs/FLUX.1-schnell" rel="nofollow"><code>black-forest-labs/FLUX.1-schnell</code></a></td></tr></tbody></table> <blockquote class="tip"><p>模型支持可以在此处请求：<a href="https://github.com/huggingface/diffusers/issues/new?template=remote-vae-pilot-feedback.yml" rel="nofollow">这里</a>。</p></blockquote> <!--[1--><h2 class="relative group"><a id="代码" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#代码"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>代码</span></h2><!--]--><!----> <blockquote class="tip"><p>从 <code>main</code> 安装 <code>diffusers</code> 以运行代码：<code>pip install git+https://github.com/huggingface/diffusers@main</code></p></blockquote> <p>一个辅助方法简化了与混合推理的交互。</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600  " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copied</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">from</span> diffusers.utils.remote_utils <span class="hljs-keyword">import</span> remote_encode<!----></pre></div><!----> <!--[2--><h3 class="relative group"><a id="基本示例" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#基本示例"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>基本示例</span></h3><!--]--><!----> <p>让我们编码一张图像，然后解码以演示。</p> <figure class="image flex flex-col items-center justify-center text-center m-0 w-full"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg"/></figure> <details><summary>代码</summary> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600  " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copied</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
<span class="hljs-keyword">from</span> diffusers.utils.remote_utils <span class="hljs-keyword">import</span> remote_decode

image = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg?download=true&quot;</span>)

latent = remote_encode(
    endpoint=<span class="hljs-string">&quot;https://ptccx55jz97f9zgo.us-east-1.aws.endpoints.huggingface.cloud/&quot;</span>,
    scaling_factor=<span class="hljs-number">0.3611</span>,
    shift_factor=<span class="hljs-number">0.1159</span>,
)

decoded = remote_decode(
    endpoint=<span class="hljs-string">&quot;https://whhx50ex1aryqvw6.us-east-1.aws.endpoints.huggingface.cloud/&quot;</span>,
    tensor=latent,
    scaling_factor=<span class="hljs-number">0.3611</span>,
    shift_factor=<span class="hljs-number">0.1159</span>,
)<!----></pre></div><!----></details> <figure class="image flex flex-col items-center justify-center text-center m-0 w-full"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/remote_vae/decoded.png"/></figure> <!--[2--><h3 class="relative group"><a id="生成" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#生成"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>生成</span></h3><!--]--><!----> <p>现在让我们看一个生成示例，我们将编码图像，生成，然后远程解码！</p> <details><summary>代码</summary> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5   text-gray-600  " title="code excerpt" type="button"><svg xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg><!----> <div class=" absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0 "><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent;"></div> Copied</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> StableDiffusionImg2ImgPip
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
<span class="hljs-keyword">from</span> diffusers.utils.remote_utils <span class="hljs-keyword">import</span> remote_decode, remote_encode

pipe = StableDiffusionImg2ImgPipeline.from_pretrained(
    <span class="hljs-string">&quot;stable-diffusion-v1-5/stable-diffusion-v1-5&quot;</span>,
    torch_dtype=torch.float16,
    variant=<span class="hljs-string">&quot;fp16&quot;</span>,
    vae=<span class="hljs-literal">None</span>,
).to(<span class="hljs-string">&quot;cuda&quot;</span>)

init_image = load_image(
    <span class="hljs-string">&quot;https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg&quot;</span>
)
init_image = init_image.resize((<span class="hljs-number">768</span>, <span class="hljs-number">512</span>))

init_latent = remote_encode(
    endpoint=<span class="hljs-string">&quot;https://qc6479g0aac6qwy9.us-east-1.aws.endpoints.huggingface.cloud/&quot;</span>,
    image=init_image,
    scaling_factor=<span class="hljs-number">0.18215</span>,
)

prompt = <span class="hljs-string">&quot;A fantasy landscape, trending on artstation&quot;</span>
latent = pipe(
    prompt=prompt,
    image=init_latent,
    strength=<span class="hljs-number">0.75</span>,
    output_type=<span class="hljs-string">&quot;latent&quot;</span>,
).images

image = remote_decode(
    endpoint=<span class="hljs-string">&quot;https://q1bj3bpq6kzilnsu.us-east-1.aws.endpoints.huggingface.cloud/&quot;</span>,
    tensor=latent,
    scaling_factor=<span class="hljs-number">0.18215</span>,
)
image.save(<span class="hljs-string">&quot;fantasy_landscape.jpg&quot;</span>)<!----></pre></div><!----></details> <figure class="image flex flex-col items-center justify-center text-center m-0 w-full"><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/remote_vae/fantasy_landscape.png"/></figure> <!--[1--><h2 class="relative group"><a id="集成" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#集成"><span><svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg><!----></span></a> <span>集成</span></h2><!--]--><!----> <ul><li><strong><a href="https://github.com/vladmandic/sdnext" rel="nofollow">SD.Next</a>:</strong> 具有直接支持混合推理功能的一体化用户界面。</li> <li><strong><a href="https://github.com/kijai/ComfyUI-HFRemoteVae" rel="nofollow">ComfyUI-HFRemoteVae</a>:</strong> 用于混合推理的 ComfyUI 节点。</li></ul> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/diffusers/blob/main/docs/source/zh/hybrid_inference/vae_encode.md" target="_blank"><svg class="mr-1" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M31,16l-7,7l-1.41-1.41L28.17,16l-5.58-5.59L24,9l7,7z"></path><path d="M1,16l7-7l1.41,1.41L3.83,16l5.58,5.59L8,23l-7-7z"></path><path d="M12.419,25.484L17.639,6.552l1.932,0.518L14.351,26.002z"></path></svg><!----> <span><span class="underline">Update</span> on GitHub</span></a><!----> <p></p><!--]--><!----><!--]--><!--]--><!--]--> <!--[-1--><!--]--><!--]-->
			
			<script>
				{
					__sveltekit_1mjeex = {
						base: "/docs/diffusers/main/zh",
						assets: "/docs/diffusers/main/zh"
					};

					const element = document.currentScript.parentElement;

					Promise.all([
						import("/docs/diffusers/main/zh/_app/immutable/entry/start.BjXnKLdN.js"),
						import("/docs/diffusers/main/zh/_app/immutable/entry/app.C861NjaH.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 9],
							data: [null,null],
							form: null,
							error: null
						});
					});
				}
			</script>