Hugging Face's logo

Buckets:
hf-doc-build
/
doc

Files
xet
hf-doc-build/doc / diffusers /v0.43.0 /zh /modular_diffusers /quickstart.html
mishig's picture
mishig
download
raw
54.6 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;快速入门&quot;,&quot;local&quot;:&quot;quickstart&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;运行流水线&quot;,&quot;local&quot;:&quot;run-a-pipeline&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;理解结构&quot;,&quot;local&quot;:&quot;understand-the-structure&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;工作流程&quot;,&quot;local&quot;:&quot;workflows&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;子块&quot;,&quot;local&quot;:&quot;sub-blocks&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;组合新的工作流&quot;,&quot;local&quot;:&quot;compose-new-workflows&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;后续步骤&quot;,&quot;local&quot;:&quot;next-steps&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
<link href="/docs/diffusers/main/en/_app/immutable/entry/start.0S0_iEoG.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/I9eooN77.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/Bupkjr9f.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/entry/app.CPdOxCbw.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/CyvZ3FS5.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/l0QnnOED.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/DnVIJfC5.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/nodes/0.DteTvmQZ.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/CoSrdQax.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/nodes/328.g2gShkQF.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/DSPDl1gz.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/D5eijdnr.js" rel="modulepreload">
<!--1x94atl--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Quickstart&quot;,&quot;local&quot;:&quot;quickstart&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Run a pipeline&quot;,&quot;local&quot;:&quot;run-a-pipeline&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Understand the structure&quot;,&quot;local&quot;:&quot;understand-the-structure&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Workflows&quot;,&quot;local&quot;:&quot;workflows&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Sub-blocks&quot;,&quot;local&quot;:&quot;sub-blocks&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Compose new workflows&quot;,&quot;local&quot;:&quot;compose-new-workflows&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Next steps&quot;,&quot;local&quot;:&quot;next-steps&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/><!---->
<link href="/docs/diffusers/main/en/_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: inline-end; margin-inline-start: 10px; display: inline-flex; position: relative; z-index: 10;"><div dir="ltr" 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 dir="auto">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="quickstart" 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="#quickstart"><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><!---->快速入门<!----></span></h1><!--]--><!----> <p>Modular Diffusers 是一个用于快速构建灵活且可定制流水线的框架。这些流水线能够超越标准 <code>DiffusionPipeline</code> 的功能。Modular Diffusers 的核心是 <a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline_blocks#diffusers.ModularPipelineBlocks">ModularPipelineBlocks</a>，它们可以与其他块组合，以适应新的工作流。这些块会被转换为 <a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline#diffusers.ModularPipeline">ModularPipeline</a>，这是一个面向用户的友好接口，用于运行生成任务。</p> <p>本指南将向您展示如何运行模块化流水线、理解其结构，并通过修改组成流水线的各个模块来进行自定义。如果您熟悉 <a href="/docs/diffusers/v0.43.0/zh/api/pipelines/overview#diffusers.DiffusionPipeline">DiffusionPipeline</a>，请先阅读 <a href="./coming_from_diffusion_pipeline">来自 DiffusionPipeline</a>，以了解其中的差异。</p> <!--[1--><h2 class="relative group"><a id="run-a-pipeline" 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="#run-a-pipeline"><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><!---->运行流水线<!----></span></h2><!--]--><!----> <p><a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline#diffusers.ModularPipeline">ModularPipeline</a> 是用于加载、运行和管理模块化流水线的主要接口。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> ModularPipeline, ComponentsManager
<span class="hljs-comment"># Use ComponentsManager to enable auto CPU offloading for memory efficiency</span>
manager = ComponentsManager()
manager.enable_auto_cpu_offload(device=<span class="hljs-string">&quot;cuda:0&quot;</span>)
pipe = ModularPipeline.from_pretrained(<span class="hljs-string">&quot;Qwen/Qwen-Image&quot;</span>, components_manager=manager)
pipe.load_components(dtype=torch.bfloat16)
image = pipe(
prompt=<span class="hljs-string">&quot;cat wizard with red hat, gandalf, lord of the rings, detailed, fantasy, cute, adorable, Pixar, Disney&quot;</span>,
).images[<span class="hljs-number">0</span>]
image<!----></pre></div><!----> <p><a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline#diffusers.ModularPipeline.from_pretrained">from_pretrained()</a> 采用延迟加载机制——它会读取配置以了解各组件的加载来源，但直到你调用 <a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline#diffusers.ModularPipeline.load_components">load_components()</a> 时才会实际加载模型权重。这让你能够控制组件的加载时机和方式。</p> <blockquote class="tip"><p><code>ComponentsManager</code> 配合 <code>enable_auto_cpu_offload</code> 可自动在 CPU 和 GPU 之间移动模型，从而降低 Qwen-Image 等大型模型的内存占用。详情请参阅 <a href="./components_manager">ComponentsManager</a> 指南。</p> <p>如果不需要 offloading，请移除 <code>components_manager</code> 参数，并使用 <code>to("cuda")</code> 手动将 pipeline 移动到目标设备上。</p></blockquote> <p>有关创建和加载管道的更多信息，请参阅<a href="https://huggingface.co/docs/diffusers/modular_diffusers/modular_pipeline#creating-a-pipeline" rel="nofollow">创建管道</a>和<a href="https://huggingface.co/docs/diffusers/modular_diffusers/modular_pipeline#loading-components" rel="nofollow">加载组件</a>指南。</p> <!--[1--><h2 class="relative group"><a id="understand-the-structure" 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="#understand-the-structure"><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><!---->理解结构<!----></span></h2><!--]--><!----> <p><a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline#diffusers.ModularPipeline">ModularPipeline</a> 由两部分组成：<strong>定义</strong>（即各个 block）和 <strong>状态</strong>（即已加载的组件和配置）。</p> <p>打印 pipeline 以查看其状态——包括各组件及其加载状态和配置。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-built_in">print</span>(pipe)<!----></pre></div><!----> <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> 已复制</div><!----></button><!----></div> <pre class=" "><!----><span class="hljs-symbol">QwenImageModularPipeline</span> {
<span class="hljs-string">&quot;_blocks_class_name&quot;</span>: <span class="hljs-string">&quot;QwenImageAutoBlocks&quot;</span>,
<span class="hljs-string">&quot;_class_name&quot;</span>: <span class="hljs-string">&quot;QwenImageModularPipeline&quot;</span>,
<span class="hljs-string">&quot;_diffusers_version&quot;</span>: <span class="hljs-string">&quot;0.37.0.dev0&quot;</span>,
<span class="hljs-string">&quot;transformer&quot;</span>: [
<span class="hljs-string">&quot;diffusers&quot;</span>,
<span class="hljs-string">&quot;QwenImageTransformer2DModel&quot;</span>,
{
<span class="hljs-string">&quot;pretrained_model_name_or_path&quot;</span>: <span class="hljs-string">&quot;Qwen/Qwen-Image&quot;</span>,
<span class="hljs-string">&quot;revision&quot;</span>: null,
<span class="hljs-string">&quot;subfolder&quot;</span>: <span class="hljs-string">&quot;transformer&quot;</span>,
<span class="hljs-string">&quot;type_hint&quot;</span>: [
<span class="hljs-string">&quot;diffusers&quot;</span>,
<span class="hljs-string">&quot;QwenImageTransformer2DModel&quot;</span>
],
<span class="hljs-string">&quot;variant&quot;</span>: null
}
],
...
}<!----></pre></div><!----> <p>通过 <code>pipe.blocks</code> 访问定义 — 这是 <a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline_blocks#diffusers.ModularPipelineBlocks">ModularPipelineBlocks</a>，用于定义流水线的工作流、输入、输出和计算逻辑。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-built_in">print</span>(pipe.blocks)<!----></pre></div><!----> <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> 已复制</div><!----></button><!----></div> <pre class=" "><!---->QwenImageAutoBlocks(
Class: SequentialPipelineBlocks
Description: Auto Modular pipeline <span class="hljs-keyword">for</span> <span class="hljs-keyword">text</span>-to-<span class="hljs-keyword">image</span>, <span class="hljs-keyword">image</span>-to-<span class="hljs-keyword">image</span>, inpainting, and controlnet tasks using QwenImage.
Supported workflows:
- <span class="hljs-string">`text2image`</span>: <span class="hljs-keyword">requires</span> <span class="hljs-string">`prompt`</span>
- <span class="hljs-string">`image2image`</span>: <span class="hljs-keyword">requires</span> <span class="hljs-string">`prompt`</span>, <span class="hljs-string">`image`</span>
- <span class="hljs-string">`inpainting`</span>: <span class="hljs-keyword">requires</span> <span class="hljs-string">`prompt`</span>, <span class="hljs-string">`mask_image`</span>, <span class="hljs-string">`image`</span>
- <span class="hljs-string">`controlnet_text2image`</span>: <span class="hljs-keyword">requires</span> <span class="hljs-string">`prompt`</span>, <span class="hljs-string">`control_image`</span>
...
Components:
text_encoder (<span class="hljs-string">`Qwen2_5_VLForConditionalGeneration`</span>)
vae (<span class="hljs-string">`AutoencoderKLQwenImage`</span>)
transformer (<span class="hljs-string">`QwenImageTransformer2DModel`</span>)
...
Sub-Blocks:
[<span class="hljs-number">0</span>] text_encoder (QwenImageAutoTextEncoderStep)
[<span class="hljs-number">1</span>] vae_encoder (QwenImageAutoVaeEncoderStep)
[<span class="hljs-number">2</span>] controlnet_vae_encoder (QwenImageOptionalControlNetVaeEncoderStep)
[<span class="hljs-number">3</span>] denoise (QwenImageAutoCoreDenoiseStep)
[<span class="hljs-number">4</span>] decode (QwenImageAutoDecodeStep)
)<!----></pre></div><!----> <p>输出返回：</p> <ul><li>支持的工作流（text2image、image2image、inpainting 等）</li> <li>它由以下子模块组成（text_encoder、vae_encoder、denoise、decode）</li></ul> <!--[2--><h3 class="relative group"><a id="workflows" 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="#workflows"><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><!---->工作流程<!----></span></h3><!--]--><!----> <p>此流水线支持多种工作流，并根据您提供的输入调整其行为。例如，如果您向流水线传入 <code>image</code>，它将执行图像到图像的工作流，而不是文本到图像的工作流。有关其底层工作原理的更多信息，请参阅 <a href="https://huggingface.co/docs/diffusers/modular_diffusers/auto_pipeline_blocks" rel="nofollow">AutoPipelineBlocks</a> 指南。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
input_image = load_image(<span class="hljs-string">&quot;https://github.com/Trgtuan10/Image_storage/blob/main/cute_cat.png?raw=true&quot;</span>)
image = pipe(
prompt=<span class="hljs-string">&quot;cat wizard with red hat, gandalf, lord of the rings, detailed, fantasy, cute, adorable, Pixar, Disney&quot;</span>,
image=input_image,
).images[<span class="hljs-number">0</span>]<!----></pre></div><!----> <p>使用 <code>get_workflow()</code> 可提取特定工作流的块。传入工作流名称（例如 <code>"image2image"</code>、<code>"inpainting"</code>、<code>"controlnet_text2image"</code>）即可仅获取与该工作流相关的块。当您想要自定义或调试特定工作流时，此方法非常有用。您可以查看 <code>pipe.blocks.available_workflows</code> 以查看所有可用的工作流。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!---->img2img_blocks = pipe.blocks.get_workflow(<span class="hljs-string">&quot;image2image&quot;</span>)<!----></pre></div><!----> <!--[2--><h3 class="relative group"><a id="sub-blocks" 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="#sub-blocks"><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><!---->子块<!----></span></h3><!--]--><!----> <p>块可以包含其他块。<code>pipe.blocks</code> 提供顶层块定义（此处为 <code>QwenImageAutoBlocks</code>），而 <code>sub_blocks</code> 则用于访问其中的较小块。</p> <p><code>QwenImageAutoBlocks</code> 由以下部分组成：<code>text_encoder</code>、<code>vae_encoder</code>、<code>controlnet_vae_encoder</code>、<code>denoise</code> 和 <code>decode</code>。</p> <p>这些子块按顺序依次执行，数据从一个块线性流向下一个块——每个块的 <code>intermediate_outputs</code> 都会作为 <code>inputs</code> 提供给下一个块。<a href="./sequential_pipeline_blocks"><code>SequentialPipelineBlocks</code></a> 就是这样工作的。</p> <p>您可以通过 <code>sub_blocks</code> 属性访问它们。<code>doc</code> 属性可用于查看任何块的完整文档，包括其输入、输出和组件。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!---->vae_encoder_block = pipe.blocks.sub_blocks[<span class="hljs-string">&quot;vae_encoder&quot;</span>]
<span class="hljs-built_in">print</span>(vae_encoder_block.doc)<!----></pre></div><!----> <p>此代码块可以转换为流水线，从而能够通过 <a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline_blocks#diffusers.ModularPipelineBlocks.init_pipeline">init_pipeline()</a> 独立运行。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!---->vae_encoder_pipe = vae_encoder_block.init_pipeline()
<span class="hljs-comment"># Reuse the VAE we already loaded, we can reuse it with update_components() method</span>
vae_encoder_pipe.update_components(vae=pipe.vae)
<span class="hljs-comment"># Run just this block</span>
image_latents = vae_encoder_pipe(image=input_image).image_latents
<span class="hljs-built_in">print</span>(image_latents.shape)<!----></pre></div><!----> <p>它复用了原始流水线中的 VAE，而不是重新加载，从而保持内存使用效率。有关详细信息，请参阅 <a href="https://huggingface.co/docs/diffusers/modular_diffusers/modular_pipeline#loading-components" rel="nofollow">加载组件</a> 指南。</p> <p>由于块是可组合的，你可以通过添加、删除或替换块来修改流水线的定义，从而创建新的工作流。在下一节中，我们将向 ControlNet 流水线添加一个 canny 边缘检测块，这样你就可以传入一张普通图像，而不是一张预先处理好的 canny 边缘图。</p> <!--[1--><h2 class="relative group"><a id="compose-new-workflows" 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="#compose-new-workflows"><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><!---->组合新的工作流<!----></span></h2><!--]--><!----> <p>让我们向 ControlNet 流水线中添加一个 canny 边缘检测块。首先，从 Hub 加载一个预构建的 canny 块（参见 <a href="https://huggingface.co/docs/diffusers/modular_diffusers/custom_blocks" rel="nofollow">构建自定义块</a> 以创建您自己的块）。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">from</span> diffusers.modular_pipelines <span class="hljs-keyword">import</span> ModularPipelineBlocks
<span class="hljs-comment"># Load a canny block from the Hub</span>
canny_block = ModularPipelineBlocks.from_pretrained(
<span class="hljs-string">&quot;diffusers-internal-dev/canny-filtering&quot;</span>,
trust_remote_code=<span class="hljs-literal">True</span>,
)
<span class="hljs-built_in">print</span>(canny_block.doc)<!----></pre></div><!----> <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> 已复制</div><!----></button><!----></div> <pre class=" "><!---->class CannyBlock
Inputs:
image (`Union[Image, ndarray]`):
Image <span class="hljs-keyword">to</span> compute canny <span class="hljs-built_in">filter</span> <span class="hljs-keyword">on</span>
low_threshold (`<span class="hljs-keyword">int</span>`, *optional*, defaults <span class="hljs-keyword">to</span> <span class="hljs-number">50</span>):
Low threshold <span class="hljs-keyword">for</span> the canny <span class="hljs-built_in">filter</span>.
high_threshold (`<span class="hljs-keyword">int</span>`, *optional*, defaults <span class="hljs-keyword">to</span> <span class="hljs-number">200</span>):
High threshold <span class="hljs-keyword">for</span> the canny <span class="hljs-built_in">filter</span>.
...
Outputs:
control_image (`PIL.Image`):
Canny <span class="hljs-keyword">map</span> <span class="hljs-keyword">for</span> <span class="hljs-built_in">input</span> image<!----></pre></div><!----> <p>使用 <code>get_workflow</code> 从 <code>QwenImageAutoBlocks</code> 中提取 ControlNet 工作流。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-comment"># Get the controlnet workflow that we want to work with</span>
blocks = pipe.blocks.get_workflow(<span class="hljs-string">&quot;controlnet_text2image&quot;</span>)
<span class="hljs-built_in">print</span>(blocks.doc)<!----></pre></div><!----> <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> 已复制</div><!----></button><!----></div> <pre class=" "><!---->class <span class="hljs-symbol">SequentialPipelineBlocks</span>
<span class="hljs-symbol">Inputs</span>:
prompt (<span class="hljs-string">`str`</span>):
<span class="hljs-symbol">The</span> prompt or prompts to guide image generation.
control_image (<span class="hljs-string">`Image`</span>):
<span class="hljs-symbol">Control</span> image for <span class="hljs-symbol">ControlNet</span> conditioning.
...<!----></pre></div><!----> <p>提取出的工作流是一个 <a href="./sequential_pipeline_blocks"><code>SequentialPipelineBlocks</code></a>，目前它需要 <code>control_image</code> 作为输入。请在开头插入 canny 块，以便该流水线能够接受普通图像作为输入。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-comment"># Insert canny at the beginning</span>
blocks.sub_blocks.insert(<span class="hljs-string">&quot;canny&quot;</span>, canny_block, <span class="hljs-number">0</span>)
<span class="hljs-comment"># Check the updated structure: CannyBlock is now listed as first sub-block</span>
<span class="hljs-built_in">print</span>(blocks)
<span class="hljs-comment"># Check the updated doc</span>
<span class="hljs-built_in">print</span>(blocks.doc)<!----></pre></div><!----> <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> 已复制</div><!----></button><!----></div> <pre class=" "><!---->class SequentialPipelineBlocks
Inputs:
image (<span class="hljs-string">`Union[Image, ndarray]`</span>):
Image <span class="hljs-keyword">to</span> compute canny filter <span class="hljs-keyword">on</span>
low_threshold (<span class="hljs-string">`int`</span>, *optional*, defaults <span class="hljs-keyword">to</span> <span class="hljs-number">50</span>):
Low threshold for the canny filter.
high_threshold (<span class="hljs-string">`int`</span>, *optional*, defaults <span class="hljs-keyword">to</span> <span class="hljs-number">200</span>):
High threshold for the canny filter.
prompt (<span class="hljs-string">`str`</span>):
The prompt <span class="hljs-literal">or</span> prompts <span class="hljs-keyword">to</span> guide image generation.
<span class="hljs-params">...</span><!----></pre></div><!----> <p>现在，pipeline 以 <code>image</code> 作为输入，而不是 <code>control_image</code>。由于序列中的各个 block 会自动共享数据，canny block 的输出（<code>control_image</code>）会流向需要它的 denoise block，而 canny block 的输入（<code>image</code>）则成为 pipeline 的输入，因为没有更早的 block 提供该数据。</p> <p>使用修改后的模块创建 pipeline，并加载 ControlNet 模型。ControlNet 不属于原始模型仓库，因此需要单独加载，并通过 <a href="/docs/diffusers/v0.43.0/zh/api/modular_diffusers/pipeline#diffusers.ModularPipeline.update_components">update_components()</a> 将其添加进去。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!---->pipeline = blocks.init_pipeline(<span class="hljs-string">&quot;Qwen/Qwen-Image&quot;</span>, components_manager=manager)
pipeline.load_components(dtype=torch.bfloat16)
<span class="hljs-comment"># Load the ControlNet model</span>
controlnet_spec = pipeline.get_component_spec(<span class="hljs-string">&quot;controlnet&quot;</span>)
controlnet_spec.pretrained_model_name_or_path = <span class="hljs-string">&quot;InstantX/Qwen-Image-ControlNet-Union&quot;</span>
controlnet = controlnet_spec.load(dtype=torch.bfloat16)
pipeline.update_components(controlnet=controlnet)<!----></pre></div><!----> <p>现在运行流水线 - canny 块会为 ControlNet 对图像进行预处理。</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> 已复制</div><!----></button><!----></div> <pre class="language-py "><!----><span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
prompt = <span class="hljs-string">&quot;cat wizard with red hat, gandalf, lord of the rings, detailed, fantasy, cute, adorable, Pixar, Disney&quot;</span>
image = load_image(<span class="hljs-string">&quot;https://github.com/Trgtuan10/Image_storage/blob/main/cute_cat.png?raw=true&quot;</span>)
output = pipeline(
prompt=prompt,
image=image,
).images[<span class="hljs-number">0</span>]
output<!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="next-steps" 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="#next-steps"><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><!---->后续步骤<!----></span></h2><!--]--><!----> <div class="flex space-x-2 items-center my-1.5 mr-8 h-7 !pl-0 -mx-3 md:mx-0"><!--[--><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd border-gray-800 bg-black dark:bg-gray-700 text-white">学习基础知识</div><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd text-gray-500 cursor-pointer opacity-90 hover:text-gray-700 dark:hover:text-gray-200 hover:shadow-sm">构建自定义块</div><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd text-gray-500 cursor-pointer opacity-90 hover:text-gray-700 dark:hover:text-gray-200 hover:shadow-sm">共享组件</div><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd text-gray-500 cursor-pointer opacity-90 hover:text-gray-700 dark:hover:text-gray-200 hover:shadow-sm">可视化界面</div><!--]--></div> <div class="language-select"><div><p>了解 Modular Diffusers 的核心构建模块：</p> <ul><li><a href="./pipeline_block">ModularPipelineBlocks</a>：用于定义流水线中一个步骤的基本单元。</li> <li><a href="./sequential_pipeline_blocks">SequentialPipelineBlocks</a>：将多个块串联起来，按顺序执行。</li> <li><a href="./auto_pipeline_blocks">AutoPipelineBlocks</a>：创建支持多种工作流的流水线。</li> <li><a href="./modular_diffusers_states">状态</a>：数据在块之间如何共享。</li></ul><!----></div><!----> <div hidden=""><p>在 <a href="./custom_blocks">构建自定义块</a> 指南中，学习如何创建具有自定义逻辑的块。</p><!----></div><!----> <div hidden=""><p>使用 <a href="./components_manager"><code>ComponentsManager</code></a> 在多个流水线之间共享模型并高效管理内存。</p><!----></div><!----> <div hidden=""><p>将模块化流水线连接到 <a href="https://github.com/cubiq/Mellon" rel="nofollow">Mellon</a>，这是一个用于构建工作流的可视化节点式界面。使用 Modular Diffusers 构建的自定义模块可直接在 Mellon 中运行，无需编写任何 UI 代码。请参阅 Mellon 指南以了解更多信息。</p><!----></div><!----><!----></div><!----> <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/en/modular_diffusers/quickstart.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_1qsdbwf = {
base: "/docs/diffusers/main/en",
assets: "/docs/diffusers/main/en"
};
const element = document.currentScript.parentElement;
Promise.all([
import("/docs/diffusers/main/en/_app/immutable/entry/start.0S0_iEoG.js"),
import("/docs/diffusers/main/en/_app/immutable/entry/app.CPdOxCbw.js")
]).then(([kit, app]) => {
kit.start(app, element, {
node_ids: [0, 328],
data: [null,null],
form: null,
error: null
});
});
}
</script>

Xet Storage Details

Size:
54.6 kB
·
Xet hash:
be5ff709730ac686c1f0d98dbcbd2397cfa94c695a1fe35f8178cf1f71c50793

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.