Hugging Face's logo

Buckets:
hf-doc-build
/
doc-dev

Files
xet
hf-doc-build/doc-dev / diffusers /pr_11686 /en /using-diffusers /ip_adapter.html
download
raw
77 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;IP-Adapter&quot;,&quot;local&quot;:&quot;ip-adapter&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Model variants&quot;,&quot;local&quot;:&quot;model-variants&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Image embeddings&quot;,&quot;local&quot;:&quot;image-embeddings&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Masking&quot;,&quot;local&quot;:&quot;masking&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Applications&quot;,&quot;local&quot;:&quot;applications&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Face models&quot;,&quot;local&quot;:&quot;face-models&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Multiple IP-Adapters&quot;,&quot;local&quot;:&quot;multiple-ip-adapters&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Instant generation&quot;,&quot;local&quot;:&quot;instant-generation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Structural control&quot;,&quot;local&quot;:&quot;structural-control&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Style and layout control&quot;,&quot;local&quot;:&quot;style-and-layout-control&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
 <link href="/docs/diffusers/pr_11686/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/entry/start.2b9667fb.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/scheduler.8c3d61f6.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/singletons.756349ae.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/index.0997d446.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/paths.8d5937da.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/entry/app.a2a6117e.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/index.da70eac4.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/nodes/0.a31d0923.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/each.e59479a4.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/nodes/294.a01749f5.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/Tip.1d9b8c37.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/CodeBlock.a9c4becf.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/getInferenceSnippets.d00e08ac.js">
 <link rel="modulepreload" href="/docs/diffusers/pr_11686/en/_app/immutable/chunks/HfOption.6ab18950.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;IP-Adapter&quot;,&quot;local&quot;:&quot;ip-adapter&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Model variants&quot;,&quot;local&quot;:&quot;model-variants&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Image embeddings&quot;,&quot;local&quot;:&quot;image-embeddings&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Masking&quot;,&quot;local&quot;:&quot;masking&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Applications&quot;,&quot;local&quot;:&quot;applications&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Face models&quot;,&quot;local&quot;:&quot;face-models&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Multiple IP-Adapters&quot;,&quot;local&quot;:&quot;multiple-ip-adapters&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Instant generation&quot;,&quot;local&quot;:&quot;instant-generation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Structural control&quot;,&quot;local&quot;:&quot;structural-control&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Style and layout control&quot;,&quot;local&quot;:&quot;style-and-layout-control&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <h1 class="relative group"><a id="ip-adapter" 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="#ip-adapter"><span><svg class="" 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>IP-Adapter</span></h1> <p data-svelte-h="svelte-lv4dxu"><a href="https://huggingface.co/papers/2308.06721" rel="nofollow">IP-Adapter</a> is a lightweight adapter designed to integrate image-based guidance with text-to-image diffusion models. The adapter uses an image encoder to extract image features that are passed to the newly added cross-attention layers in the UNet and fine-tuned. The original UNet model and the existing cross-attention layers corresponding to text features is frozen. Decoupling the cross-attention for image and text features enables more fine-grained and controllable generation.</p> <p data-svelte-h="svelte-w1ah1v">IP-Adapter files are typically ~100MBs because they only contain the image embeddings. This means you need to load a model first, and then load the IP-Adapter with <a href="/docs/diffusers/pr_11686/en/api/loaders/ip_adapter#diffusers.loaders.IPAdapterMixin.load_ip_adapter">load_ip_adapter()</a>.</p> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-1s8spky">IP-Adapters are available to many models such as <a href="../api/pipelines/flux#ip-adapter">Flux</a> and <a href="../api/pipelines/stable_diffusion/stable_diffusion_3">Stable Diffusion 3</a>, and more. The examples in this guide use Stable Diffusion and Stable Diffusion XL.</p></div> <p data-svelte-h="svelte-1fj8tz0">Use the <a href="/docs/diffusers/pr_11686/en/api/loaders/ip_adapter#diffusers.loaders.IPAdapterMixin.set_ip_adapter_scale">set_ip_adapter_scale()</a> parameter to scale the influence of the IP-Adapter during generation. A value of <code>1.0</code> means the model is only conditioned on the image prompt, and <code>0.5</code> typically produces balanced results between the text and image prompt.</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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> AutoPipelineForText2Image
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
pipeline = AutoPipelineForText2Image.from_pretrained(
 <span class="hljs-string">&quot;stabilityai/stable-diffusion-xl-base-1.0&quot;</span>,
 torch_dtype=torch.float16
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;sdxl_models&quot;</span>,
 weight_name=<span class="hljs-string">&quot;ip-adapter_sdxl.bin&quot;</span>
)
pipeline.set_ip_adapter_scale(<span class="hljs-number">0.8</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1wker05">Pass an image to <code>ip_adapter_image</code> along with a text prompt to generate an image.</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 class="" 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=""><!-- HTML_TAG_START -->image = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_diner.png&quot;</span>)
pipeline(
 prompt=<span class="hljs-string">&quot;a polar bear sitting in a chair drinking a milkshake&quot;</span>,
 ip_adapter_image=image,
 negative_prompt=<span class="hljs-string">&quot;deformed, ugly, wrong proportion, low res, bad anatomy, worst quality, low quality&quot;</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-295lls"><figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_diner.png" width="400" alt="IP-Adapter image"> <figcaption style="text-align: center;">IP-Adapter image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_diner_2.png" width="400" alt="generated image"> <figcaption style="text-align: center;">generated image</figcaption></figure></div> <p data-svelte-h="svelte-12n8oag">Take a look at the examples below to learn how to use IP-Adapter for other tasks.</p> <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">image-to-image </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">inpainting </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">video </div></div> <div class="language-select"><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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> AutoPipelineForImage2Image
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
pipeline = AutoPipelineForImage2Image.from_pretrained(
 <span class="hljs-string">&quot;stabilityai/stable-diffusion-xl-base-1.0&quot;</span>,
 torch_dtype=torch.float16
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;sdxl_models&quot;</span>,
 weight_name=<span class="hljs-string">&quot;ip-adapter_sdxl.bin&quot;</span>
)
pipeline.set_ip_adapter_scale(<span class="hljs-number">0.8</span>)
image = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_bear_1.png&quot;</span>)
ip_image = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_gummy.png&quot;</span>)
pipeline(
 prompt=<span class="hljs-string">&quot;best quality, high quality&quot;</span>,
 image=image,
 ip_adapter_image=ip_image,
 strength=<span class="hljs-number">0.5</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-143vd3e"><figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_bear_1.png" width="300" alt="input image"> <figcaption style="text-align: center;">input image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_gummy.png" width="300" alt="IP-Adapter image"> <figcaption style="text-align: center;">IP-Adapter image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_bear_3.png" width="300" alt="generated image"> <figcaption style="text-align: center;">generated image</figcaption></figure></div> </div> <h2 class="relative group"><a id="model-variants" 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="#model-variants"><span><svg class="" 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>Model variants</span></h2> <p data-svelte-h="svelte-9uieuk">There are two variants of IP-Adapter, Plus and FaceID. The Plus variant uses patch embeddings and the ViT-H image encoder. FaceID variant uses face embeddings generated from InsightFace.</p> <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">IP-Adapter Plus </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">IP-Adapter FaceID </div></div> <div class="language-select"><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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> CLIPVisionModelWithProjection, AutoPipelineForText2Image
image_encoder = CLIPVisionModelWithProjection.from_pretrained(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;models/image_encoder&quot;</span>,
 torch_dtype=torch.float16
)
pipeline = AutoPipelineForText2Image.from_pretrained(
 <span class="hljs-string">&quot;stabilityai/stable-diffusion-xl-base-1.0&quot;</span>,
 image_encoder=image_encoder,
 torch_dtype=torch.float16
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;sdxl_models&quot;</span>,
 weight_name=<span class="hljs-string">&quot;ip-adapter-plus_sdxl_vit-h.safetensors&quot;</span>
)<!-- HTML_TAG_END --></pre></div> </div> <h2 class="relative group"><a id="image-embeddings" 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="#image-embeddings"><span><svg class="" 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>Image embeddings</span></h2> <p data-svelte-h="svelte-hpizde">The <code>prepare_ip_adapter_image_embeds</code> generates image embeddings you can reuse if you’re running the pipeline multiple times because you have more than one image. Loading and encoding multiple images each time you use the pipeline can be inefficient. Precomputing the image embeddings ahead of time, saving them to disk, and loading them when you need them is more efficient.</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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> AutoPipelineForText2Image
pipeline = AutoPipelineForImage2Image.from_pretrained(
 <span class="hljs-string">&quot;stabilityai/stable-diffusion-xl-base-1.0&quot;</span>,
 torch_dtype=torch.float16
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
image_embeds = pipeline.prepare_ip_adapter_image_embeds(
 ip_adapter_image=image,
 ip_adapter_image_embeds=<span class="hljs-literal">None</span>,
 device=<span class="hljs-string">&quot;cuda&quot;</span>,
 num_images_per_prompt=<span class="hljs-number">1</span>,
 do_classifier_free_guidance=<span class="hljs-literal">True</span>,
)
torch.save(image_embeds, <span class="hljs-string">&quot;image_embeds.ipadpt&quot;</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-17yemw9">Reload the image embeddings by passing them to the <code>ip_adapter_image_embeds</code> parameter. Set <code>image_encoder_folder</code> to <code>None</code> because you don’t need the image encoder anymore to generate the image embeddings.</p> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-15eph5s">You can also load image embeddings from other sources such as ComfyUI.</p></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 class="" 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=""><!-- HTML_TAG_START -->pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;sdxl_models&quot;</span>,
 image_encoder_folder=<span class="hljs-literal">None</span>,
 weight_name=<span class="hljs-string">&quot;ip-adapter_sdxl.bin&quot;</span>
)
pipeline.set_ip_adapter_scale(<span class="hljs-number">0.8</span>)
image_embeds = torch.load(<span class="hljs-string">&quot;image_embeds.ipadpt&quot;</span>)
pipeline(
 prompt=<span class="hljs-string">&quot;a polar bear sitting in a chair drinking a milkshake&quot;</span>,
 ip_adapter_image_embeds=image_embeds,
 negative_prompt=<span class="hljs-string">&quot;deformed, ugly, wrong proportion, low res, bad anatomy, worst quality, low quality&quot;</span>,
 num_inference_steps=<span class="hljs-number">100</span>,
 generator=generator,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="masking" 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="#masking"><span><svg class="" 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>Masking</span></h2> <p data-svelte-h="svelte-1btur7m">Binary masking enables assigning an IP-Adapter image to a specific area of the output image, making it useful for composing multiple IP-Adapter images. Each IP-Adapter image requires a binary mask.</p> <p data-svelte-h="svelte-1ipm5aa">Load the <a href="/docs/diffusers/pr_11686/en/api/loaders/ip_adapter#diffusers.image_processor.IPAdapterMaskProcessor">IPAdapterMaskProcessor</a> to preprocess the image masks. For the best results, provide the output <code>height</code> and <code>width</code> to ensure masks with different aspect ratios are appropriately sized. If the input masks already match the aspect ratio of the generated image, you don’t need to set the <code>height</code> and <code>width</code>.</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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> AutoPipelineForText2Image
<span class="hljs-keyword">from</span> diffusers.image_processor <span class="hljs-keyword">import</span> IPAdapterMaskProcessor
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
pipeline = AutoPipelineForImage2Image.from_pretrained(
 <span class="hljs-string">&quot;stabilityai/stable-diffusion-xl-base-1.0&quot;</span>,
 torch_dtype=torch.float16
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
mask1 = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_mask_mask1.png&quot;</span>)
mask2 = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_mask_mask2.png&quot;</span>)
processor = IPAdapterMaskProcessor()
masks = processor.preprocess([mask1, mask2], height=<span class="hljs-number">1024</span>, width=<span class="hljs-number">1024</span>)<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-8iu0me"><figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ip_mask_mask1.png" width="200" alt="mask 1"> <figcaption style="text-align: center;">mask 1</figcaption></figure> <figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ip_mask_mask2.png" width="200" alt="mask 2"> <figcaption style="text-align: center;">mask 2</figcaption></figure></div> <p data-svelte-h="svelte-bmwit0">Provide both the IP-Adapter images and their scales as a list. Pass the preprocessed masks to <code>cross_attention_kwargs</code> in the pipeline.</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 class="" 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=""><!-- HTML_TAG_START -->face_image1 = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_mask_girl1.png&quot;</span>)
face_image2 = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_mask_girl2.png&quot;</span>)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;sdxl_models&quot;</span>,
 weight_name=[<span class="hljs-string">&quot;ip-adapter-plus-face_sdxl_vit-h.safetensors&quot;</span>]
)
pipeline.set_ip_adapter_scale([[<span class="hljs-number">0.7</span>, <span class="hljs-number">0.7</span>]])
ip_images = [[face_image1, face_image2]]
masks = [masks.reshape(<span class="hljs-number">1</span>, masks.shape[<span class="hljs-number">0</span>], masks.shape[<span class="hljs-number">2</span>], masks.shape[<span class="hljs-number">3</span>])]
pipeline(
 prompt=<span class="hljs-string">&quot;2 girls&quot;</span>,
 ip_adapter_image=ip_images,
 negative_prompt=<span class="hljs-string">&quot;monochrome, lowres, bad anatomy, worst quality, low quality&quot;</span>,
 cross_attention_kwargs={<span class="hljs-string">&quot;ip_adapter_masks&quot;</span>: masks}
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; flex-direction: column; gap: 10px;" data-svelte-h="svelte-1cx2gkd"><div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;"><figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ip_mask_girl1.png" width="400" alt="IP-Adapter image 1"> <figcaption style="text-align: center;">IP-Adapter image 1</figcaption></figure> <figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ip_mask_girl2.png" width="400" alt="IP-Adapter image 2"> <figcaption style="text-align: center;">IP-Adapter image 2</figcaption></figure></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;"><figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_attention_mask_result_seed_0.png" width="400" alt="Generated image with mask"> <figcaption style="text-align: center;">generated with mask</figcaption></figure> <figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_no_attention_mask_result_seed_0.png" width="400" alt="Generated image without mask"> <figcaption style="text-align: center;">generated without mask</figcaption></figure></div></div> <h2 class="relative group"><a id="applications" 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="#applications"><span><svg class="" 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>Applications</span></h2> <p data-svelte-h="svelte-1j6fwoj">The section below covers some popular applications of IP-Adapter.</p> <h3 class="relative group"><a id="face-models" 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="#face-models"><span><svg class="" 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>Face models</span></h3> <p data-svelte-h="svelte-jdscb4">Face generation and preserving its details can be challenging. To help generate more accurate faces, there are checkpoints specifically conditioned on images of cropped faces. You can find the face models in the <a href="https://huggingface.co/h94/IP-Adapter" rel="nofollow">h94/IP-Adapter</a> repository or the <a href="https://huggingface.co/h94/IP-Adapter-FaceID" rel="nofollow">h94/IP-Adapter-FaceID</a> repository. The FaceID checkpoints use the FaceID embeddings from <a href="https://github.com/deepinsight/insightface" rel="nofollow">InsightFace</a> instead of CLIP image embeddings.</p> <p data-svelte-h="svelte-7kw9cj">We recommend using the <a href="/docs/diffusers/pr_11686/en/api/schedulers/ddim#diffusers.DDIMScheduler">DDIMScheduler</a> or <a href="/docs/diffusers/pr_11686/en/api/schedulers/euler#diffusers.EulerDiscreteScheduler">EulerDiscreteScheduler</a> for face models.</p> <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">h94/IP-Adapter </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">h94/IP-Adapter-FaceID </div></div> <div class="language-select"><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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> StableDiffusionPipeline, DDIMScheduler
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
pipeline = StableDiffusionPipeline.from_pretrained(
 <span class="hljs-string">&quot;stable-diffusion-v1-5/stable-diffusion-v1-5&quot;</span>,
 torch_dtype=torch.float16,
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;models&quot;</span>,
weight_name=<span class="hljs-string">&quot;ip-adapter-full-face_sd15.bin&quot;</span>
)
pipeline.set_ip_adapter_scale(<span class="hljs-number">0.5</span>)
image = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_einstein_base.png&quot;</span>)
pipeline(
 prompt=<span class="hljs-string">&quot;A photo of Einstein as a chef, wearing an apron, cooking in a French restaurant&quot;</span>,
 ip_adapter_image=image,
 negative_prompt=<span class="hljs-string">&quot;lowres, bad anatomy, worst quality, low quality&quot;</span>,
 num_inference_steps=<span class="hljs-number">100</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-1sqhdfz"><figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_einstein_base.png" width="400" alt="IP-Adapter image"> <figcaption style="text-align: center;">IP-Adapter image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_einstein.png" width="400" alt="generated image"> <figcaption style="text-align: center;">generated image</figcaption></figure></div> </div> <h3 class="relative group"><a id="multiple-ip-adapters" 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="#multiple-ip-adapters"><span><svg class="" 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>Multiple IP-Adapters</span></h3> <p data-svelte-h="svelte-8t9z4d">Combine multiple IP-Adapters to generate images in more diverse styles. For example, you can use IP-Adapter Face to generate consistent faces and characters and IP-Adapter Plus to generate those faces in specific styles.</p> <p data-svelte-h="svelte-1rlgp6a">Load an image encoder with <a href="https://huggingface.co/docs/transformers/main/en/model_doc/clip#transformers.CLIPVisionModelWithProjection" rel="nofollow">CLIPVisionModelWithProjection</a>.</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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> AutoPipelineForText2Image, DDIMScheduler
<span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> CLIPVisionModelWithProjection
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
image_encoder = CLIPVisionModelWithProjection.from_pretrained(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;models/image_encoder&quot;</span>,
 torch_dtype=torch.float16,
)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-12sbcxj">Load a base model, scheduler and the following IP-Adapters.</p> <ul data-svelte-h="svelte-14lybvh"><li><a href="https://huggingface.co/h94/IP-Adapter#ip-adapter-for-sdxl-10" rel="nofollow">ip-adapter-plus_sdxl_vit-h</a> uses patch embeddings and a ViT-H image encoder</li> <li><a href="https://huggingface.co/h94/IP-Adapter#ip-adapter-for-sdxl-10" rel="nofollow">ip-adapter-plus-face_sdxl_vit-h</a> uses patch embeddings and a ViT-H image encoder but it is conditioned on images of cropped faces</li></ul> <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 class="" 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=""><!-- HTML_TAG_START -->pipeline = AutoPipelineForText2Image.from_pretrained(
 <span class="hljs-string">&quot;stabilityai/stable-diffusion-xl-base-1.0&quot;</span>,
 torch_dtype=torch.float16,
 image_encoder=image_encoder,
)
pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;sdxl_models&quot;</span>,
 weight_name=[<span class="hljs-string">&quot;ip-adapter-plus_sdxl_vit-h.safetensors&quot;</span>, <span class="hljs-string">&quot;ip-adapter-plus-face_sdxl_vit-h.safetensors&quot;</span>]
)
pipeline.set_ip_adapter_scale([<span class="hljs-number">0.7</span>, <span class="hljs-number">0.3</span>])
<span class="hljs-comment"># enable_model_cpu_offload to reduce memory usage</span>
pipeline.enable_model_cpu_offload()<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1lm8xpa">Load an image and a folder containing images of a certain style to apply.</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 class="" 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=""><!-- HTML_TAG_START -->face_image = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/women_input.png&quot;</span>)
style_folder = <span class="hljs-string">&quot;https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/style_ziggy&quot;</span>
style_images = [load_image(<span class="hljs-string">f&quot;<span class="hljs-subst">{style_folder}</span>/img<span class="hljs-subst">{i}</span>.png&quot;</span>) <span class="hljs-keyword">for</span> i <span class="hljs-keyword">in</span> <span class="hljs-built_in">range</span>(<span class="hljs-number">10</span>)]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-1fqw2fw"><figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/women_input.png" width="400" alt="Face image"> <figcaption style="text-align: center;">face image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ip_style_grid.png" width="400" alt="Style images"> <figcaption style="text-align: center;">style images</figcaption></figure></div> <p data-svelte-h="svelte-1f9ayt4">Pass style and face images as a list to <code>ip_adapter_image</code>.</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 class="" 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=""><!-- HTML_TAG_START -->generator = torch.Generator(device=<span class="hljs-string">&quot;cpu&quot;</span>).manual_seed(<span class="hljs-number">0</span>)
pipeline(
 prompt=<span class="hljs-string">&quot;wonderwoman&quot;</span>,
 ip_adapter_image=[style_images, face_image],
 negative_prompt=<span class="hljs-string">&quot;monochrome, lowres, bad anatomy, worst quality, low quality&quot;</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; justify-content: center;" data-svelte-h="svelte-1vyi3xu"><figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ip_multi_out.png" width="400" alt="Generated image"> <figcaption style="text-align: center;">generated image</figcaption></figure></div> <h3 class="relative group"><a id="instant-generation" 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="#instant-generation"><span><svg class="" 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>Instant generation</span></h3> <p data-svelte-h="svelte-gtun4"><a href="../api/pipelines/latent_consistency_models">Latent Consistency Models (LCM)</a> can generate images 4 steps or less, unlike other diffusion models which require a lot more steps, making it feel “instantaneous”. IP-Adapters are compatible with LCM models to instantly generate images.</p> <p data-svelte-h="svelte-fn159h">Load the IP-Adapter weights and load the LoRA weights with <a href="/docs/diffusers/pr_11686/en/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights">load_lora_weights()</a>.</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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> DiffusionPipeline, LCMScheduler
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
pipeline = DiffusionPipeline.from_pretrained(
 <span class="hljs-string">&quot;sd-dreambooth-library/herge-style&quot;</span>,
 torch_dtype=torch.float16
)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;models&quot;</span>,
 weight_name=<span class="hljs-string">&quot;ip-adapter_sd15.bin&quot;</span>
)
pipeline.load_lora_weights(<span class="hljs-string">&quot;latent-consistency/lcm-lora-sdv1-5&quot;</span>)
pipeline.scheduler = LCMScheduler.from_config(pipeline.scheduler.config)
<span class="hljs-comment"># enable_model_cpu_offload to reduce memory usage</span>
pipeline.enable_model_cpu_offload()<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-9f9j6q">Try using a lower IP-Adapter scale to condition generation more on the style you want to apply and remember to use the special token in your prompt to trigger its generation.</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 class="" 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=""><!-- HTML_TAG_START -->pipeline.set_ip_adapter_scale(<span class="hljs-number">0.4</span>)
prompt = <span class="hljs-string">&quot;herge_style woman in armor, best quality, high quality&quot;</span>
ip_adapter_image = load_image(<span class="hljs-string">&quot;https://user-images.githubusercontent.com/24734142/266492875-2d50d223-8475-44f0-a7c6-08b51cb53572.png&quot;</span>)
pipeline(
 prompt=prompt,
 ip_adapter_image=ip_adapter_image,
 num_inference_steps=<span class="hljs-number">4</span>,
 guidance_scale=<span class="hljs-number">1</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; justify-content: center;" data-svelte-h="svelte-uhx8ym"><figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ip_adapter_herge.png" width="400" alt="Generated image"> <figcaption style="text-align: center;">generated image</figcaption></figure></div> <h3 class="relative group"><a id="structural-control" 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="#structural-control"><span><svg class="" 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>Structural control</span></h3> <p data-svelte-h="svelte-f8td34">For structural control, combine IP-Adapter with <a href="../api/pipelines/controlnet">ControlNet</a> conditioned on depth maps, edge maps, pose estimations, and more.</p> <p data-svelte-h="svelte-pbj9a5">The example below loads a <a href="/docs/diffusers/pr_11686/en/api/models/controlnet#diffusers.ControlNetModel">ControlNetModel</a> checkpoint conditioned on depth maps and combines it with a IP-Adapter.</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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> StableDiffusionControlNetPipeline, ControlNetModel
controlnet = ControlNetModel.from_pretrained(
 <span class="hljs-string">&quot;lllyasviel/control_v11f1p_sd15_depth&quot;</span>,
 torch_dtype=torch.float16
)
pipeline = StableDiffusionControlNetPipeline.from_pretrained(
 <span class="hljs-string">&quot;stable-diffusion-v1-5/stable-diffusion-v1-5&quot;</span>,
 controlnet=controlnet,
 torch_dtype=torch.float16
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;models&quot;</span>,
 weight_name=<span class="hljs-string">&quot;ip-adapter_sd15.bin&quot;</span>
)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1grcvlc">Pass the depth map and IP-Adapter image to the pipeline.</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 class="" 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=""><!-- HTML_TAG_START -->pipeline(
 prompt=<span class="hljs-string">&quot;best quality, high quality&quot;</span>,
 image=depth_map,
 ip_adapter_image=ip_adapter_image,
 negative_prompt=<span class="hljs-string">&quot;monochrome, lowres, bad anatomy, worst quality, low quality&quot;</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-11joxfd"><figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/statue.png" width="300" alt="IP-Adapter image"> <figcaption style="text-align: center;">IP-Adapter image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/depth.png" width="300" alt="Depth map"> <figcaption style="text-align: center;">depth map</figcaption></figure> <figure><img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ipa-controlnet-out.png" width="300" alt="Generated image"> <figcaption style="text-align: center;">generated image</figcaption></figure></div> <h3 class="relative group"><a id="style-and-layout-control" 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="#style-and-layout-control"><span><svg class="" 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>Style and layout control</span></h3> <p data-svelte-h="svelte-znjjtu">For style and layout control, combine IP-Adapter with <a href="https://huggingface.co/papers/2404.02733" rel="nofollow">InstantStyle</a>. InstantStyle separates <em>style</em> (color, texture, overall feel) and <em>content</em> from each other. It only applies the style in style-specific blocks of the model to prevent it from distorting other areas of an image. This generates images with stronger and more consistent styles and better control over the layout.</p> <p data-svelte-h="svelte-vd8f3b">The IP-Adapter is only activated for specific parts of the model. Use the <a href="/docs/diffusers/pr_11686/en/api/loaders/ip_adapter#diffusers.loaders.IPAdapterMixin.set_ip_adapter_scale">set_ip_adapter_scale()</a> method to scale the influence of the IP-Adapter in different layers. The example below activates the IP-Adapter in the second layer of the models down <code>block_2</code> and up <code>block_0</code>. Down <code>block_2</code> is where the IP-Adapter injects layout information and up <code>block_0</code> is where style is injected.</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 class="" 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=""><!-- HTML_TAG_START --><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> AutoPipelineForText2Image
<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> load_image
pipeline = AutoPipelineForText2Image.from_pretrained(
 <span class="hljs-string">&quot;stabilityai/stable-diffusion-xl-base-1.0&quot;</span>,
 torch_dtype=torch.float16
).to(<span class="hljs-string">&quot;cuda&quot;</span>)
pipeline.load_ip_adapter(
 <span class="hljs-string">&quot;h94/IP-Adapter&quot;</span>,
 subfolder=<span class="hljs-string">&quot;sdxl_models&quot;</span>,
 weight_name=<span class="hljs-string">&quot;ip-adapter_sdxl.bin&quot;</span>
)
scale = {
 <span class="hljs-string">&quot;down&quot;</span>: {<span class="hljs-string">&quot;block_2&quot;</span>: [<span class="hljs-number">0.0</span>, <span class="hljs-number">1.0</span>]},
 <span class="hljs-string">&quot;up&quot;</span>: {<span class="hljs-string">&quot;block_0&quot;</span>: [<span class="hljs-number">0.0</span>, <span class="hljs-number">1.0</span>, <span class="hljs-number">0.0</span>]},
}
pipeline.set_ip_adapter_scale(scale)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-a2xkl9">Load the style image and generate an image.</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 class="" 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=""><!-- HTML_TAG_START -->style_image = load_image(<span class="hljs-string">&quot;https://huggingface.co/datasets/huggingface/documentation-images/resolve/0052a70beed5bf71b92610a43a52df6d286cd5f3/diffusers/rabbit.jpg&quot;</span>)
pipeline(
 prompt=<span class="hljs-string">&quot;a cat, masterpiece, best quality, high quality&quot;</span>,
 ip_adapter_image=style_image,
 negative_prompt=<span class="hljs-string">&quot;text, watermark, lowres, low quality, worst quality, deformed, glitch, low contrast, noisy, saturation, blurry&quot;</span>,
 guidance_scale=<span class="hljs-number">5</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-1i6rw5e"><figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/0052a70beed5bf71b92610a43a52df6d286cd5f3/diffusers/rabbit.jpg" width="400" alt="Style image"> <figcaption style="text-align: center;">style image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/datasets/cat_style_layout.png" width="400" alt="Generated image"> <figcaption style="text-align: center;">generated image</figcaption></figure></div> <p data-svelte-h="svelte-1z0qe12">You can also insert the IP-Adapter in all the model layers. This tends to generate images that focus more on the image prompt and may reduce the diversity of generated images. Only activate the IP-Adapter in up <code>block_0</code> or the style layer.</p> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-1wnfs9b">You don’t need to specify all the layers in the <code>scale</code> dictionary. Layers not included are set to 0, which means the IP-Adapter is disabled.</p></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 class="" 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=""><!-- HTML_TAG_START -->scale = {
 <span class="hljs-string">&quot;up&quot;</span>: {<span class="hljs-string">&quot;block_0&quot;</span>: [<span class="hljs-number">0.0</span>, <span class="hljs-number">1.0</span>, <span class="hljs-number">0.0</span>]},
}
pipeline.set_ip_adapter_scale(scale)
pipeline(
 prompt=<span class="hljs-string">&quot;a cat, masterpiece, best quality, high quality&quot;</span>,
 ip_adapter_image=style_image,
 negative_prompt=<span class="hljs-string">&quot;text, watermark, lowres, low quality, worst quality, deformed, glitch, low contrast, noisy, saturation, blurry&quot;</span>,
 guidance_scale=<span class="hljs-number">5</span>,
).images[<span class="hljs-number">0</span>]<!-- HTML_TAG_END --></pre></div> <div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;" data-svelte-h="svelte-sniway"><figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/datasets/cat_style_only.png" width="400" alt="Generated image (style only)"> <figcaption style="text-align: center;">style-layer generated image</figcaption></figure> <figure><img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/datasets/cat_ip_adapter.png" width="400" alt="Generated image (IP-Adapter only)"> <figcaption style="text-align: center;">all layers generated image</figcaption></figure></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/using-diffusers/ip_adapter.md" target="_blank"><span data-svelte-h="svelte-1kd6by1">&lt;</span> <span data-svelte-h="svelte-x0xyl0">&gt;</span> <span data-svelte-h="svelte-1dajgef"><span class="underline ml-1.5">Update</span> on GitHub</span></a> <p></p>
<script>
 {
 __sveltekit_o82a48 = {
 assets: "/docs/diffusers/pr_11686/en",
 base: "/docs/diffusers/pr_11686/en",
 env: {}
 };
 const element = document.currentScript.parentElement;
 const data = [null,null];
 Promise.all([
 import("/docs/diffusers/pr_11686/en/_app/immutable/entry/start.2b9667fb.js"),
 import("/docs/diffusers/pr_11686/en/_app/immutable/entry/app.a2a6117e.js")
 ]).then(([kit, app]) => {
 kit.start(app, element, {
 node_ids: [0, 294],
 data,
 form: null,
 error: null
 });
 });
 }
 </script>

Xet Storage Details

Size:
77 kB
·
Xet hash:
b3bed0762be1c460c3b3c638096ef5cb81277d01d011b78c1369b1b409b3ff30

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