diffusers/v0.19.0/en/using-diffusers/weighted_prompts.html

<meta charset="utf-8" /><meta http-equiv="content-security-policy" content=""><meta name="hf:doc:metadata" content="{&quot;local&quot;:&quot;weighting-prompts&quot;,&quot;sections&quot;:[{&quot;local&quot;:&quot;how-to-do-promptweighting-in-diffusers&quot;,&quot;title&quot;:&quot;How to do prompt-weighting in Diffusers&quot;}],&quot;title&quot;:&quot;Weighting prompts&quot;}" data-svelte="svelte-1phssyn">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/assets/pages/__layout.svelte-hf-doc-builder.css">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/start-hf-doc-builder.js">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/chunks/vendor-hf-doc-builder.js">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/chunks/paths-hf-doc-builder.js">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/pages/__layout.svelte-hf-doc-builder.js">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/pages/using-diffusers/weighted_prompts.mdx-hf-doc-builder.js">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/chunks/IconCopyLink-hf-doc-builder.js">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/chunks/CodeBlock-hf-doc-builder.js">
	<link rel="modulepreload" href="/docs/diffusers/v0.19.0/en/_app/chunks/DocNotebookDropdown-hf-doc-builder.js"> 






<h1 class="relative group"><a id="weighting-prompts" 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="#weighting-prompts"><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>Weighting prompts
	</span></h1>



<div class="flex space-x-1 absolute z-10 right-0 top-0">
	<div class="relative colab-dropdown ">
	<button class="  " type="button">
		
				<img alt="Open In Colab" class="!m-0" src="https://colab.research.google.com/assets/colab-badge.svg">
			</button>
	
	
	
	</div>
	<div class="relative colab-dropdown ">
	<button class="  " type="button">
		
				<img alt="Open In Studio Lab" class="!m-0" src="https://studiolab.sagemaker.aws/studiolab.svg">
			</button>
	
	
	
	</div></div>
<p>Text-guided diffusion models generate images based on a given text prompt. The text prompt
can include multiple concepts that the model should generate and it’s often desirable to weight
certain parts of the prompt more or less. </p>
<p>Diffusion models work by conditioning the cross attention layers of the diffusion model with contextualized text embeddings (see the <a href="../stable-diffusion">Stable Diffusion Guide for more information</a>).
Thus a simple way to emphasize (or de-emphasize) certain parts of the prompt is by increasing or reducing the scale of the text embedding vector that corresponds to the relevant part of the prompt.
This is called “prompt-weighting” and has been a highly demanded feature by the community (see issue <a href="https://github.com/huggingface/diffusers/issues/2431" rel="nofollow">here</a>).</p>
<h2 class="relative group"><a id="how-to-do-promptweighting-in-diffusers" 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="#how-to-do-promptweighting-in-diffusers"><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>How to do prompt-weighting in Diffusers
	</span></h2>

<p>We believe the role of <code>diffusers</code> is to be a toolbox that provides essential features that enable other projects, such as <a href="https://github.com/invoke-ai/InvokeAI" rel="nofollow">InvokeAI</a> or <a href="https://github.com/abhishekkrthakur/diffuzers" rel="nofollow">diffuzers</a>, to build powerful UIs. In order to support arbitrary methods to manipulate prompts, <code>diffusers</code> exposes a <a href="https://huggingface.co/docs/diffusers/v0.14.0/en/api/pipelines/stable_diffusion/text2img#diffusers.StableDiffusionPipeline.__call__.prompt_embeds" rel="nofollow"><code>prompt_embeds</code></a> function argument to many pipelines such as <a href="/docs/diffusers/v0.19.0/en/api/pipelines/stable_diffusion/text2img#diffusers.StableDiffusionPipeline">StableDiffusionPipeline</a>, allowing to directly pass the “prompt-weighted”/scaled text embeddings to the pipeline.</p>
<p>The <a href="https://github.com/damian0815/compel" rel="nofollow">compel library</a> provides an easy way to emphasize or de-emphasize portions of the prompt for you. We strongly recommend it instead of preparing the embeddings yourself.</p>
<p>Let’s look at a simple example. Imagine you want to generate an image of <code>&quot;a red cat playing with a ball&quot;</code> as
follows:</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><!-- HTML_TAG_START --><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> StableDiffusionPipeline, UniPCMultistepScheduler

pipe = StableDiffusionPipeline.from_pretrained(<span class="hljs-string">&quot;CompVis/stable-diffusion-v1-4&quot;</span>)
pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)

prompt = <span class="hljs-string">&quot;a red cat playing with a ball&quot;</span>

generator = torch.Generator(device=<span class="hljs-string">&quot;cpu&quot;</span>).manual_seed(<span class="hljs-number">33</span>)

image = pipe(prompt, generator=generator, num_inference_steps=<span class="hljs-number">20</span>).images[<span class="hljs-number">0</span>]
image<!-- HTML_TAG_END --></pre></div>
<p>This gives you:</p>
<p><img src="https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/compel/forest_0.png" alt="img"></p>
<p>As you can see, there is no “ball” in the image. Let’s emphasize this part!</p>
<p>For this we should install the <code>compel</code> library:</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><!-- HTML_TAG_START -->pip <span class="hljs-keyword">install</span> compel<!-- HTML_TAG_END --></pre></div>
<p>and then create a <code>Compel</code> object:</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><!-- HTML_TAG_START --><span class="hljs-keyword">from</span> compel <span class="hljs-keyword">import</span> Compel

compel_proc = Compel(tokenizer=pipe.tokenizer, text_encoder=pipe.text_encoder)<!-- HTML_TAG_END --></pre></div>
<p>Now we emphasize the part “ball” with the <code>&quot;++&quot;</code> syntax:</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><!-- HTML_TAG_START -->prompt = <span class="hljs-string">&quot;a red cat playing with a ball++&quot;</span><!-- HTML_TAG_END --></pre></div>
<p>and instead of passing this to the pipeline directly, we have to process it using <code>compel_proc</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><!-- HTML_TAG_START -->prompt_embeds = compel_proc(prompt)<!-- HTML_TAG_END --></pre></div>
<p>Now we can pass <code>prompt_embeds</code> directly 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><!-- HTML_TAG_START -->generator = torch.Generator(device=<span class="hljs-string">&quot;cpu&quot;</span>).manual_seed(<span class="hljs-number">33</span>)

images = pipe(prompt_embeds=prompt_embeds, generator=generator, num_inference_steps=<span class="hljs-number">20</span>).images[<span class="hljs-number">0</span>]
image<!-- HTML_TAG_END --></pre></div>
<p>We now get the following image which has a “ball”!</p>
<p><img src="https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/compel/forest_1.png" alt="img"></p>
<p>Similarly, we de-emphasize parts of the sentence by using the <code>--</code> suffix for words, feel free to give it
a try!</p>
<p>If your favorite pipeline does not have a <code>prompt_embeds</code> input, please make sure to open an issue, the
diffusers team tries to be as responsive as possible.</p>
<p>Compel 1.1.6 adds a utility class to simplify using textual inversions.  Instantiate a <code>DiffusersTextualInversionManager</code> and pass it to Compel init:</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><!-- HTML_TAG_START --><span class="hljs-attr">textual_inversion_manager</span> = DiffusersTextualInversionManager(pipe)
<span class="hljs-attr">compel</span> = Compel(
    <span class="hljs-attr">tokenizer</span>=pipe.tokenizer,
    <span class="hljs-attr">text_encoder</span>=pipe.text_encoder,
    <span class="hljs-attr">textual_inversion_manager</span>=textual_inversion_manager)<!-- HTML_TAG_END --></pre></div>
<p>Also, please check out the documentation of the <a href="https://github.com/damian0815/compel" rel="nofollow">compel</a> library for
more information.</p>


		<script type="module" data-hydrate="1kkxvlp">
		import { start } from "/docs/diffusers/v0.19.0/en/_app/start-hf-doc-builder.js";
		start({
			target: document.querySelector('[data-hydrate="1kkxvlp"]').parentNode,
			paths: {"base":"/docs/diffusers/v0.19.0/en","assets":"/docs/diffusers/v0.19.0/en"},
			session: {},
			route: false,
			spa: false,
			trailing_slash: "never",
			hydrate: {
				status: 200,
				error: null,
				nodes: [
					import("/docs/diffusers/v0.19.0/en/_app/pages/__layout.svelte-hf-doc-builder.js"),
						import("/docs/diffusers/v0.19.0/en/_app/pages/using-diffusers/weighted_prompts.mdx-hf-doc-builder.js")
				],
				params: {}
			}
		});
	</script>