Buckets:

hf-doc-build
/

doc-dev

hf-doc-build/doc-dev / diffusers /pr_14192 /en /_app /immutable /nodes /161.C9G9eaN_.js

HuggingFaceDocBuilder's picture

HuggingFaceDocBuilder

26 days ago

53.4 kB

	import"../chunks/DsnmJJEf.js";import{i as S,h as F,H as s,a as i,D as n,E as D,s as L}from"../chunks/BtE7mKSK.js";import{p as A,o as K,s as e,f as h,a as d,b as O,c as a,d as f,r as t,n as W}from"../chunks/jDjavuwI.js";import{E as N}from"../chunks/SrSJA0zO.js";const $='{"title":"HunyuanImage2.1","local":"hunyuanimage21","sections":[{"title":"HunyuanImage-2.1","local":"hunyuanimage-21","sections":[],"depth":2},{"title":"HunyuanImage-2.1-Distilled","local":"hunyuanimage-21-distilled","sections":[],"depth":2},{"title":"HunyuanImagePipeline","local":"diffusers.HunyuanImagePipeline","sections":[],"depth":2},{"title":"HunyuanImageRefinerPipeline","local":"diffusers.HunyuanImageRefinerPipeline","sections":[],"depth":2},{"title":"HunyuanImagePipelineOutput","local":"diffusers.pipelines.hunyuan_image.pipeline_output.HunyuanImagePipelineOutput","sections":[],"depth":2}],"depth":1}';var ee=f('<meta name="hf:doc:metadata"/>'),R=f("<p>Examples:</p> <!>",1),ne=f('<p></p> <!> <p>HunyuanImage-2.1 is a 17B text-to-image model that is capable of generating 2K (2048 x 2048) resolution images</p> <p>HunyuanImage-2.1 comes in the following variants:</p> <table><thead><tr><th align="center">model type</th><th align="center">model id</th></tr></thead><tbody><tr><td align="center">HunyuanImage-2.1</td><td align="center"><a href="https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Diffusers" rel="nofollow">hunyuanvideo-community/HunyuanImage-2.1-Diffusers</a></td></tr><tr><td align="center">HunyuanImage-2.1-Distilled</td><td align="center"><a href="https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Distilled-Diffusers" rel="nofollow">hunyuanvideo-community/HunyuanImage-2.1-Distilled-Diffusers</a></td></tr><tr><td align="center">HunyuanImage-2.1-Refiner</td><td align="center"><a href="https://huggingface.co/hunyuanvideo-community/HunyuanImage-2.1-Refiner-Diffusers" rel="nofollow">hunyuanvideo-community/HunyuanImage-2.1-Refiner-Diffusers</a></td></tr></tbody></table> <blockquote><p>[!TIP][Caching](../../optimization/cache) may also speed up inference by storing and reusing intermediate outputs.</p></blockquote> <!> <p>HunyuanImage-2.1 applies <a href="https://huggingface.co/papers/2410.02416" rel="nofollow">Adaptive Projected Guidance (APG)</a> combined with Classifier-Free Guidance (CFG) in the denoising loop. <code>HunyuanImagePipeline</code> has a <code>guider</code> component (read more about <a href="../../using-diffusers/guiders">Guider</a>) and does not take a <code>guidance_scale</code> parameter at runtime. To change guider-related parameters, e.g., <code>guidance_scale</code>, you can update the <code>guider</code> configuration instead.</p> <!> <p>You can inspect the <code>guider</code> object:</p> <!> <p>To update the guider with a different configuration, use the <code>new()</code> method. For example, to generate an image with <code>guidance_scale=5.0</code> while keeping all other default guidance parameters:</p> <!> <!> <p>use <code>distilled_guidance_scale</code> with the guidance-distilled checkpoint,</p> <!> <!> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!> <p>The HunyuanImage pipeline for text-to-image generation.</p> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!> <p>Function invoked when calling the pipeline for generation.</p> <!></div> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!></div></div> <!> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!> <p>The HunyuanImage pipeline for text-to-image generation.</p> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!> <p>Function invoked when calling the pipeline for generation.</p> <!></div> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!></div></div> <!> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!> <p>Output class for HunyuanImage pipelines.</p></div> <!> <p></p>',1);function ie(C,V){A(V,!1),K(()=>{new URLSearchParams(window.location.search).get("fw")}),S();var y=ne();F("16ehmo8",o=>{var l=ee();L(l,"content",$),d(o,l)});var b=e(h(y),2);s(b,{title:"HunyuanImage2.1",local:"hunyuanimage21",headingTag:"h1"});var w=e(b,10);s(w,{title:"HunyuanImage-2.1",local:"hunyuanimage-21",headingTag:"h2"});var M=e(w,4);i(M,{code:"aW1wb3J0JTIwdG9yY2glMEFmcm9tJTIwZGlmZnVzZXJzJTIwaW1wb3J0JTIwSHVueXVhbkltYWdlUGlwZWxpbmUlMEElMEFwaXBlJTIwJTNEJTIwSHVueXVhbkltYWdlUGlwZWxpbmUuZnJvbV9wcmV0cmFpbmVkKCUwQSUyMCUyMCUyMCUyMCUyMmh1bnl1YW52aWRlby1jb21tdW5pdHklMkZIdW55dWFuSW1hZ2UtMi4xLURpZmZ1c2VycyUyMiUyQyUyMCUwQSUyMCUyMCUyMCUyMHRvcmNoX2R0eXBlJTNEdG9yY2guYmZsb2F0MTYlMEEpJTBBcGlwZSUyMCUzRCUyMHBpcGUudG8oJTIyY3VkYSUyMik=",highlighted:`<span class="hljs-keyword">import</span> torch
	<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> HunyuanImagePipeline

	pipe = HunyuanImagePipeline.from_pretrained(
	<span class="hljs-string">"hunyuanvideo-community/HunyuanImage-2.1-Diffusers"</span>,
	torch_dtype=torch.bfloat16
	)
	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)`,lang:"python",wrap:!1});var I=e(M,4);i(I,{code:"cGlwZS5ndWlkZXIlMEE=",highlighted:`<span class="hljs-meta">>>> </span>pipe.guider
	AdaptiveProjectedMixGuidance {
	<span class="hljs-string">"_class_name"</span>: <span class="hljs-string">"AdaptiveProjectedMixGuidance"</span>,
	<span class="hljs-string">"_diffusers_version"</span>: <span class="hljs-string">"0.36.0.dev0"</span>,
	<span class="hljs-string">"adaptive_projected_guidance_momentum"</span>: -<span class="hljs-number">0.5</span>,
	<span class="hljs-string">"adaptive_projected_guidance_rescale"</span>: <span class="hljs-number">10.0</span>,
	<span class="hljs-string">"adaptive_projected_guidance_scale"</span>: <span class="hljs-number">10.0</span>,
	<span class="hljs-string">"adaptive_projected_guidance_start_step"</span>: <span class="hljs-number">5</span>,
	<span class="hljs-string">"enabled"</span>: true,
	<span class="hljs-string">"eta"</span>: <span class="hljs-number">0.0</span>,
	<span class="hljs-string">"guidance_rescale"</span>: <span class="hljs-number">0.0</span>,
	<span class="hljs-string">"guidance_scale"</span>: <span class="hljs-number">3.5</span>,
	<span class="hljs-string">"start"</span>: <span class="hljs-number">0.0</span>,
	<span class="hljs-string">"stop"</span>: <span class="hljs-number">1.0</span>,
	<span class="hljs-string">"use_original_formulation"</span>: false
	}

	State:
	step: <span class="hljs-literal">None</span>
	num_inference_steps: <span class="hljs-literal">None</span>
	timestep: <span class="hljs-literal">None</span>
	count_prepared: <span class="hljs-number">0</span>
	enabled: <span class="hljs-literal">True</span>
	num_conditions: <span class="hljs-number">2</span>
	momentum_buffer: <span class="hljs-literal">None</span>
	is_apg_enabled: <span class="hljs-literal">False</span>
	is_cfg_enabled: <span class="hljs-literal">True</span>`,lang:"py",wrap:!1});var v=e(I,4);i(v,{code:"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",highlighted:`<span class="hljs-keyword">import</span> torch
	<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> HunyuanImagePipeline

	pipe = HunyuanImagePipeline.from_pretrained(
	<span class="hljs-string">"hunyuanvideo-community/HunyuanImage-2.1-Diffusers"</span>,
	torch_dtype=torch.bfloat16
	)
	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)

	<span class="hljs-comment"># Update the guider configuration</span>
	pipe.guider = pipe.guider.new(guidance_scale=<span class="hljs-number">5.0</span>)

	prompt = (
	<span class="hljs-string">"A cute, cartoon-style anthropomorphic penguin plush toy with fluffy fur, standing in a painting studio, "</span>
	<span class="hljs-string">"wearing a red knitted scarf and a red beret with the word 'Tencent' on it, holding a paintbrush with a "</span>
	<span class="hljs-string">"focused expression as it paints an oil painting of the Mona Lisa, rendered in a photorealistic photographic style."</span>
	)

	image = pipe(
	prompt=prompt,
	num_inference_steps=<span class="hljs-number">50</span>,
	height=<span class="hljs-number">2048</span>,
	width=<span class="hljs-number">2048</span>,
	).images[<span class="hljs-number">0</span>]
	image.save(<span class="hljs-string">"image.png"</span>)`,lang:"py",wrap:!1});var T=e(v,2);s(T,{title:"HunyuanImage-2.1-Distilled",local:"hunyuanimage-21-distilled",headingTag:"h2"});var U=e(T,4);i(U,{code:"aW1wb3J0JTIwdG9yY2glMEFmcm9tJTIwZGlmZnVzZXJzJTIwaW1wb3J0JTIwSHVueXVhbkltYWdlUGlwZWxpbmUlMEFwaXBlJTIwJTNEJTIwSHVueXVhbkltYWdlUGlwZWxpbmUuZnJvbV9wcmV0cmFpbmVkKCUyMmh1bnl1YW52aWRlby1jb21tdW5pdHklMkZIdW55dWFuSW1hZ2UtMi4xLURpc3RpbGxlZC1EaWZmdXNlcnMlMjIlMkMlMjB0b3JjaF9kdHlwZSUzRHRvcmNoLmJmbG9hdDE2KSUwQXBpcGUlMjAlM0QlMjBwaXBlLnRvKCUyMmN1ZGElMjIpJTBBJTBBcHJvbXB0JTIwJTNEJTIwKCUwQSUyMCUyMCUyMCUyMCUyMkElMjBjdXRlJTJDJTIwY2FydG9vbi1zdHlsZSUyMGFudGhyb3BvbW9ycGhpYyUyMHBlbmd1aW4lMjBwbHVzaCUyMHRveSUyMHdpdGglMjBmbHVmZnklMjBmdXIlMkMlMjBzdGFuZGluZyUyMGluJTIwYSUyMHBhaW50aW5nJTIwc3R1ZGlvJTJDJTIwJTIyJTBBJTIwJTIwJTIwJTIwJTIyd2VhcmluZyUyMGElMjByZWQlMjBrbml0dGVkJTIwc2NhcmYlMjBhbmQlMjBhJTIwcmVkJTIwYmVyZXQlMjB3aXRoJTIwdGhlJTIwd29yZCUyMCdUZW5jZW50JyUyMG9uJTIwaXQlMkMlMjBob2xkaW5nJTIwYSUyMHBhaW50YnJ1c2glMjB3aXRoJTIwYSUyMCUyMiUwQSUyMCUyMCUyMCUyMCUyMmZvY3VzZWQlMjBleHByZXNzaW9uJTIwYXMlMjBpdCUyMHBhaW50cyUyMGFuJTIwb2lsJTIwcGFpbnRpbmclMjBvZiUyMHRoZSUyME1vbmElMjBMaXNhJTJDJTIwcmVuZGVyZWQlMjBpbiUyMGElMjBwaG90b3JlYWxpc3RpYyUyMHBob3RvZ3JhcGhpYyUyMHN0eWxlLiUyMiUwQSklMEElMEFvdXQlMjAlM0QlMjBwaXBlKCUwQSUyMCUyMCUyMCUyMHByb21wdCUyQyUwQSUyMCUyMCUyMCUyMG51bV9pbmZlcmVuY2Vfc3RlcHMlM0Q4JTJDJTBBJTIwJTIwJTIwJTIwZGlzdGlsbGVkX2d1aWRhbmNlX3NjYWxlJTNEMy4yNSUyQyUwQSUyMCUyMCUyMCUyMGhlaWdodCUzRDIwNDglMkMlMEElMjAlMjAlMjAlMjB3aWR0aCUzRDIwNDglMkMlMEElMjAlMjAlMjAlMjBnZW5lcmF0b3IlM0RnZW5lcmF0b3IlMkMlMEEpLmltYWdlcyU1QjAlNUQlMEE=",highlighted:`<span class="hljs-keyword">import</span> torch
	<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> HunyuanImagePipeline
	pipe = HunyuanImagePipeline.from_pretrained(<span class="hljs-string">"hunyuanvideo-community/HunyuanImage-2.1-Distilled-Diffusers"</span>, torch_dtype=torch.bfloat16)
	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)

	prompt = (
	<span class="hljs-string">"A cute, cartoon-style anthropomorphic penguin plush toy with fluffy fur, standing in a painting studio, "</span>
	<span class="hljs-string">"wearing a red knitted scarf and a red beret with the word 'Tencent' on it, holding a paintbrush with a "</span>
	<span class="hljs-string">"focused expression as it paints an oil painting of the Mona Lisa, rendered in a photorealistic photographic style."</span>
	)

	out = pipe(
	prompt,
	num_inference_steps=<span class="hljs-number">8</span>,
	distilled_guidance_scale=<span class="hljs-number">3.25</span>,
	height=<span class="hljs-number">2048</span>,
	width=<span class="hljs-number">2048</span>,
	generator=generator,
	).images[<span class="hljs-number">0</span>]
	`,lang:"py",wrap:!1});var j=e(U,2);s(j,{title:"HunyuanImagePipeline",local:"diffusers.HunyuanImagePipeline",headingTag:"h2"});var p=e(j,2),k=a(p);n(k,{name:"class diffusers.HunyuanImagePipeline",anchor:"diffusers.HunyuanImagePipeline",source:"https://github.com/huggingface/diffusers/blob/vr_14192/src/diffusers/pipelines/hunyuan_image/pipeline_hunyuanimage.py#L160",parameters:[{name:"scheduler",val:": FlowMatchEulerDiscreteScheduler"},{name:"vae",val:": AutoencoderKLHunyuanImage"},{name:"text_encoder",val:": Qwen2_5_VLForConditionalGeneration"},{name:"tokenizer",val:": Qwen2Tokenizer"},{name:"text_encoder_2",val:": T5EncoderModel"},{name:"tokenizer_2",val:": ByT5Tokenizer"},{name:"transformer",val:": HunyuanImageTransformer2DModel"},{name:"guider",val:": AdaptiveProjectedMixGuidance \| None = None"},{name:"ocr_guider",val:": AdaptiveProjectedMixGuidance \| None = None"}],parametersDescription:[{anchor:"diffusers.HunyuanImagePipeline.transformer",description:`<strong>transformer</strong> (<a href="/docs/diffusers/pr_14192/en/api/models/hunyuanimage_transformer_2d#diffusers.HunyuanImageTransformer2DModel">HunyuanImageTransformer2DModel</a>) —
	Conditional Transformer (MMDiT) architecture to denoise the encoded image latents.`,name:"transformer"},{anchor:"diffusers.HunyuanImagePipeline.scheduler",description:`<strong>scheduler</strong> (<a href="/docs/diffusers/pr_14192/en/api/schedulers/flow_match_euler_discrete#diffusers.FlowMatchEulerDiscreteScheduler">FlowMatchEulerDiscreteScheduler</a>) —
	A scheduler to be used in combination with <code>transformer</code> to denoise the encoded image latents.`,name:"scheduler"},{anchor:"diffusers.HunyuanImagePipeline.vae",description:`<strong>vae</strong> (<a href="/docs/diffusers/pr_14192/en/api/models/autoencoder_kl_hunyuanimage#diffusers.AutoencoderKLHunyuanImage">AutoencoderKLHunyuanImage</a>) —
	Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.`,name:"vae"},{anchor:"diffusers.HunyuanImagePipeline.text_encoder",description:`<strong>text_encoder</strong> (<code>Qwen2.5-VL-7B-Instruct</code>) —
	<a href="https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct" rel="nofollow">Qwen2.5-VL-7B-Instruct</a>, specifically the
	<a href="https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct" rel="nofollow">Qwen2.5-VL-7B-Instruct</a> variant.`,name:"text_encoder"},{anchor:"diffusers.HunyuanImagePipeline.tokenizer",description:"<strong>tokenizer</strong> (<code>Qwen2Tokenizer</code>) — Tokenizer of class [Qwen2Tokenizer].",name:"tokenizer"},{anchor:"diffusers.HunyuanImagePipeline.text_encoder_2",description:`<strong>text_encoder_2</strong> (<code>T5EncoderModel</code>) —
	<a href="https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5EncoderModel" rel="nofollow">T5EncoderModel</a>
	variant.`,name:"text_encoder_2"},{anchor:"diffusers.HunyuanImagePipeline.tokenizer_2",description:"<strong>tokenizer_2</strong> (<code>ByT5Tokenizer</code>) — Tokenizer of class [ByT5Tokenizer]",name:"tokenizer_2"},{anchor:"diffusers.HunyuanImagePipeline.guider",description:`<strong>guider</strong> (<code>AdaptiveProjectedMixGuidance</code>) —
	[AdaptiveProjectedMixGuidance]to be used to guide the image generation.`,name:"guider"},{anchor:"diffusers.HunyuanImagePipeline.ocr_guider",description:`<strong>ocr_guider</strong> (<code>AdaptiveProjectedMixGuidance</code>, <em>optional</em>) —
	[AdaptiveProjectedMixGuidance] to be used to guide the image generation when text rendering is needed.`,name:"ocr_guider"}]});var c=e(k,4),J=a(c);n(J,{name:"__call__",anchor:"diffusers.HunyuanImagePipeline.__call__",source:"https://github.com/huggingface/diffusers/blob/vr_14192/src/diffusers/pipelines/hunyuan_image/pipeline_hunyuanimage.py#L504",parameters:[{name:"prompt",val:": str \| list[str] = None"},{name:"negative_prompt",val:": str \| list[str] = None"},{name:"height",val:": int \| None = None"},{name:"width",val:": int \| None = None"},{name:"num_inference_steps",val:": int = 50"},{name:"distilled_guidance_scale",val:": float \| None = 3.25"},{name:"sigmas",val:": list[float] \| None = None"},{name:"num_images_per_prompt",val:": int = 1"},{name:"generator",val:": torch.Generator \| list[torch.Generator] \| None = None"},{name:"latents",val:": torch.Tensor \| None = None"},{name:"prompt_embeds",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_mask",val:": torch.Tensor \| None = None"},{name:"negative_prompt_embeds",val:": torch.Tensor \| None = None"},{name:"negative_prompt_embeds_mask",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_2",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_mask_2",val:": torch.Tensor \| None = None"},{name:"negative_prompt_embeds_2",val:": torch.Tensor \| None = None"},{name:"negative_prompt_embeds_mask_2",val:": torch.Tensor \| None = None"},{name:"output_type",val:": str \| None = 'pil'"},{name:"return_dict",val:": bool = True"},{name:"attention_kwargs",val:": dict[str, Any] \| None = None"},{name:"callback_on_step_end",val:": Callable[[int, int], None] \| None = None"},{name:"callback_on_step_end_tensor_inputs",val:": list[str] = ['latents']"}],parametersDescription:[{anchor:"diffusers.HunyuanImagePipeline.__call__.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>list[str]</code>, <em>optional</em>) —
	The prompt or prompts to guide the image generation. If not defined, one has to pass <code>prompt_embeds</code>.
	instead.`,name:"prompt"},{anchor:"diffusers.HunyuanImagePipeline.__call__.negative_prompt",description:`<strong>negative_prompt</strong> (<code>str</code> or <code>list[str]</code>, <em>optional</em>) —
	The prompt or prompts not to guide the image generation. If not defined and negative_prompt_embeds is
	not provided, will use an empty negative prompt. Ignored when not using guidance. ).`,name:"negative_prompt"},{anchor:"diffusers.HunyuanImagePipeline.__call__.height",description:`<strong>height</strong> (<code>int</code>, <em>optional</em>, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
	The height in pixels of the generated image. This is set to 1024 by default for the best results.`,name:"height"},{anchor:"diffusers.HunyuanImagePipeline.__call__.width",description:`<strong>width</strong> (<code>int</code>, <em>optional</em>, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
	The width in pixels of the generated image. This is set to 1024 by default for the best results.`,name:"width"},{anchor:"diffusers.HunyuanImagePipeline.__call__.num_inference_steps",description:`<strong>num_inference_steps</strong> (<code>int</code>, <em>optional</em>, defaults to 50) —
	The number of denoising steps. More denoising steps usually lead to a higher quality image at the
	expense of slower inference.`,name:"num_inference_steps"},{anchor:"diffusers.HunyuanImagePipeline.__call__.sigmas",description:`<strong>sigmas</strong> (<code>list[float]</code>, <em>optional</em>) —
	Custom sigmas to use for the denoising process with schedulers which support a <code>sigmas</code> argument in
	their <code>set_timesteps</code> method. If not defined, the default behavior when <code>num_inference_steps</code> is passed
	will be used.`,name:"sigmas"},{anchor:"diffusers.HunyuanImagePipeline.__call__.distilled_guidance_scale",description:`<strong>distilled_guidance_scale</strong> (<code>float</code>, <em>optional</em>, defaults to None) —
	A guidance scale value for guidance distilled models. Unlike the traditional classifier-free guidance
	where the guidance scale is applied during inference through noise prediction rescaling, guidance
	distilled models take the guidance scale directly as an input parameter during forward pass. Guidance
	is enabled by setting <code>distilled_guidance_scale > 1</code>. Higher guidance scale encourages to generate
	images that are closely linked to the text <code>prompt</code>, usually at the expense of lower image quality. For
	guidance distilled models, this parameter is required. For non-distilled models, this parameter will be
	ignored.`,name:"distilled_guidance_scale"},{anchor:"diffusers.HunyuanImagePipeline.__call__.num_images_per_prompt",description:`<strong>num_images_per_prompt</strong> (<code>int</code>, <em>optional</em>, defaults to 1) —
	The number of images to generate per prompt.`,name:"num_images_per_prompt"},{anchor:"diffusers.HunyuanImagePipeline.__call__.generator",description:`<strong>generator</strong> (<code>torch.Generator</code> or <code>list[torch.Generator]</code>, <em>optional</em>) —
	One or a list of <a href="https://pytorch.org/docs/stable/generated/torch.Generator.html" rel="nofollow">torch generator(s)</a>
	to make generation deterministic.`,name:"generator"},{anchor:"diffusers.HunyuanImagePipeline.__call__.latents",description:`<strong>latents</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
	generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
	tensor will be generated by sampling using the supplied random <code>generator</code>.`,name:"latents"},{anchor:"diffusers.HunyuanImagePipeline.__call__.prompt_embeds",description:`<strong>prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings. Can be used to easily tweak text inputs, <em>e.g.</em> prompt weighting. If not
	provided, text embeddings will be generated from <code>prompt</code> input argument.`,name:"prompt_embeds"},{anchor:"diffusers.HunyuanImagePipeline.__call__.prompt_embeds_mask",description:`<strong>prompt_embeds_mask</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings mask. Can be used to easily tweak text inputs, <em>e.g.</em> prompt weighting.
	If not provided, text embeddings mask will be generated from <code>prompt</code> input argument.`,name:"prompt_embeds_mask"},{anchor:"diffusers.HunyuanImagePipeline.__call__.prompt_embeds_2",description:`<strong>prompt_embeds_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings for ocr. Can be used to easily tweak text inputs, <em>e.g.</em> prompt
	weighting. If not provided, text embeddings for ocr will be generated from <code>prompt</code> input argument.`,name:"prompt_embeds_2"},{anchor:"diffusers.HunyuanImagePipeline.__call__.prompt_embeds_mask_2",description:`<strong>prompt_embeds_mask_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings mask for ocr. Can be used to easily tweak text inputs, <em>e.g.</em> prompt
	weighting. If not provided, text embeddings mask for ocr will be generated from <code>prompt</code> input
	argument.`,name:"prompt_embeds_mask_2"},{anchor:"diffusers.HunyuanImagePipeline.__call__.negative_prompt_embeds",description:`<strong>negative_prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated negative text embeddings. Can be used to easily tweak text inputs, <em>e.g.</em> prompt
	weighting. If not provided, negative_prompt_embeds will be generated from <code>negative_prompt</code> input
	argument.`,name:"negative_prompt_embeds"},{anchor:"diffusers.HunyuanImagePipeline.__call__.negative_prompt_embeds_mask",description:`<strong>negative_prompt_embeds_mask</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated negative text embeddings mask. Can be used to easily tweak text inputs, <em>e.g.</em> prompt
	weighting. If not provided, negative text embeddings mask will be generated from <code>negative_prompt</code>
	input argument.`,name:"negative_prompt_embeds_mask"},{anchor:"diffusers.HunyuanImagePipeline.__call__.negative_prompt_embeds_2",description:`<strong>negative_prompt_embeds_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated negative text embeddings for ocr. Can be used to easily tweak text inputs, <em>e.g.</em> prompt
	weighting. If not provided, negative text embeddings for ocr will be generated from <code>negative_prompt</code>
	input argument.`,name:"negative_prompt_embeds_2"},{anchor:"diffusers.HunyuanImagePipeline.__call__.negative_prompt_embeds_mask_2",description:`<strong>negative_prompt_embeds_mask_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated negative text embeddings mask for ocr. Can be used to easily tweak text inputs, <em>e.g.</em>
	prompt weighting. If not provided, negative text embeddings mask for ocr will be generated from
	<code>negative_prompt</code> input argument.`,name:"negative_prompt_embeds_mask_2"},{anchor:"diffusers.HunyuanImagePipeline.__call__.output_type",description:`<strong>output_type</strong> (<code>str</code>, <em>optional</em>, defaults to <code>"pil"</code>) —
	The output format of the generate image. Choose between
	<a href="https://pillow.readthedocs.io/en/stable/" rel="nofollow">PIL</a>: <code>PIL.Image.Image</code> or <code>np.array</code>.`,name:"output_type"},{anchor:"diffusers.HunyuanImagePipeline.__call__.return_dict",description:`<strong>return_dict</strong> (<code>bool</code>, <em>optional</em>, defaults to <code>True</code>) —
	Whether or not to return a <code>~pipelines.qwenimage.QwenImagePipelineOutput</code> instead of a plain tuple.`,name:"return_dict"},{anchor:"diffusers.HunyuanImagePipeline.__call__.attention_kwargs",description:`<strong>attention_kwargs</strong> (<code>dict</code>, <em>optional</em>) —
	A kwargs dictionary that if specified is passed along to the <code>AttentionProcessor</code> as defined under
	<code>self.processor</code> in
	<a href="https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py" rel="nofollow">diffusers.models.attention_processor</a>.`,name:"attention_kwargs"},{anchor:"diffusers.HunyuanImagePipeline.__call__.callback_on_step_end",description:`<strong>callback_on_step_end</strong> (<code>Callable</code>, <em>optional</em>) —
	A function that calls at the end of each denoising steps during the inference. The function is called
	with the following arguments: <code>callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)</code>. <code>callback_kwargs</code> will include a list of all tensors as specified by
	<code>callback_on_step_end_tensor_inputs</code>.`,name:"callback_on_step_end"},{anchor:"diffusers.HunyuanImagePipeline.__call__.callback_on_step_end_tensor_inputs",description:`<strong>callback_on_step_end_tensor_inputs</strong> (<code>List</code>, <em>optional</em>) —
	The list of tensor inputs for the <code>callback_on_step_end</code> function. The tensors specified in the list
	will be passed as <code>callback_kwargs</code> argument. You will only be able to include variables listed in the
	<code>._callback_tensor_inputs</code> attribute of your pipeline class.`,name:"callback_on_step_end_tensor_inputs"}],returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>~pipelines.hunyuan_image.HunyuanImagePipelineOutput</code> if <code>return_dict</code> is True, otherwise a <code>tuple</code>. When
	returning a tuple, the first element is a list with the generated images.</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>~pipelines.hunyuan_image.HunyuanImagePipelineOutput</code> or <code>tuple</code></p>
	`});var E=e(J,4);N(E,{anchor:"diffusers.HunyuanImagePipeline.__call__.example",children:(o,l)=>{var r=R(),_=e(h(r),2);i(_,{code:"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",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">import</span> torch
	<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> HunyuanImagePipeline

	<span class="hljs-meta">>>> </span>pipe = HunyuanImagePipeline.from_pretrained(
	<span class="hljs-meta">... </span> <span class="hljs-string">"hunyuanvideo-community/HunyuanImage-2.1-Diffusers"</span>, torch_dtype=torch.bfloat16
	<span class="hljs-meta">... </span>)
	<span class="hljs-meta">>>> </span>pipe.to(<span class="hljs-string">"cuda"</span>)
	<span class="hljs-meta">>>> </span>prompt = <span class="hljs-string">"A cat holding a sign that says hello world"</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># Depending on the variant being used, the pipeline call will slightly vary.</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># Refer to the pipeline documentation for more details.</span>
	<span class="hljs-meta">>>> </span>image = pipe(prompt, negative_prompt=<span class="hljs-string">""</span>, num_inference_steps=<span class="hljs-number">50</span>).images[<span class="hljs-number">0</span>]
	<span class="hljs-meta">>>> </span>image.save(<span class="hljs-string">"hunyuanimage.png"</span>)`,lang:"py",wrap:!1}),d(o,r)},$$slots:{default:!0}}),t(c);var H=e(c,2),q=a(H);n(q,{name:"encode_prompt",anchor:"diffusers.HunyuanImagePipeline.encode_prompt",source:"https://github.com/huggingface/diffusers/blob/vr_14192/src/diffusers/pipelines/hunyuan_image/pipeline_hunyuanimage.py#L296",parameters:[{name:"prompt",val:": str \| list[str]"},{name:"device",val:": torch.device \| None = None"},{name:"batch_size",val:": int = 1"},{name:"num_images_per_prompt",val:": int = 1"},{name:"prompt_embeds",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_mask",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_2",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_mask_2",val:": torch.Tensor \| None = None"}],parametersDescription:[{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>list[str]</code>, <em>optional</em>) —
	prompt to be encoded`,name:"prompt"},{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.device",description:`<strong>device</strong> — (<code>torch.device</code>):
	torch device`,name:"device"},{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.batch_size",description:`<strong>batch_size</strong> (<code>int</code>) —
	batch size of prompts, defaults to 1`,name:"batch_size"},{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.num_images_per_prompt",description:`<strong>num_images_per_prompt</strong> (<code>int</code>) —
	number of images that should be generated per prompt`,name:"num_images_per_prompt"},{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.prompt_embeds",description:`<strong>prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings. If not provided, text embeddings will be generated from <code>prompt</code> input
	argument.`,name:"prompt_embeds"},{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.prompt_embeds_mask",description:`<strong>prompt_embeds_mask</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text mask. If not provided, text mask will be generated from <code>prompt</code> input argument.`,name:"prompt_embeds_mask"},{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.prompt_embeds_2",description:`<strong>prompt_embeds_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated glyph text embeddings from ByT5. If not provided, will be generated from <code>prompt</code> input
	argument using self.tokenizer_2 and self.text_encoder_2.`,name:"prompt_embeds_2"},{anchor:"diffusers.HunyuanImagePipeline.encode_prompt.prompt_embeds_mask_2",description:`<strong>prompt_embeds_mask_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated glyph text mask from ByT5. If not provided, will be generated from <code>prompt</code> input
	argument using self.tokenizer_2 and self.text_encoder_2.`,name:"prompt_embeds_mask_2"}]}),t(H),t(p);var x=e(p,2);s(x,{title:"HunyuanImageRefinerPipeline",local:"diffusers.HunyuanImageRefinerPipeline",headingTag:"h2"});var m=e(x,2),B=a(m);n(B,{name:"class diffusers.HunyuanImageRefinerPipeline",anchor:"diffusers.HunyuanImageRefinerPipeline",source:"https://github.com/huggingface/diffusers/blob/vr_14192/src/diffusers/pipelines/hunyuan_image/pipeline_hunyuanimage_refiner.py#L138",parameters:[{name:"scheduler",val:": FlowMatchEulerDiscreteScheduler"},{name:"vae",val:": AutoencoderKLHunyuanImageRefiner"},{name:"text_encoder",val:": Qwen2_5_VLForConditionalGeneration"},{name:"tokenizer",val:": Qwen2Tokenizer"},{name:"transformer",val:": HunyuanImageTransformer2DModel"},{name:"guider",val:": AdaptiveProjectedMixGuidance \| None = None"}],parametersDescription:[{anchor:"diffusers.HunyuanImageRefinerPipeline.transformer",description:`<strong>transformer</strong> (<a href="/docs/diffusers/pr_14192/en/api/models/hunyuanimage_transformer_2d#diffusers.HunyuanImageTransformer2DModel">HunyuanImageTransformer2DModel</a>) —
	Conditional Transformer (MMDiT) architecture to denoise the encoded image latents.`,name:"transformer"},{anchor:"diffusers.HunyuanImageRefinerPipeline.scheduler",description:`<strong>scheduler</strong> (<a href="/docs/diffusers/pr_14192/en/api/schedulers/flow_match_euler_discrete#diffusers.FlowMatchEulerDiscreteScheduler">FlowMatchEulerDiscreteScheduler</a>) —
	A scheduler to be used in combination with <code>transformer</code> to denoise the encoded image latents.`,name:"scheduler"},{anchor:"diffusers.HunyuanImageRefinerPipeline.vae",description:`<strong>vae</strong> (<a href="/docs/diffusers/pr_14192/en/api/models/autoencoder_kl_hunyuanimage_refiner#diffusers.AutoencoderKLHunyuanImageRefiner">AutoencoderKLHunyuanImageRefiner</a>) —
	Variational Auto-Encoder (VAE) Model to encode and decode images to and from latent representations.`,name:"vae"},{anchor:"diffusers.HunyuanImageRefinerPipeline.text_encoder",description:`<strong>text_encoder</strong> (<code>Qwen2.5-VL-7B-Instruct</code>) —
	<a href="https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct" rel="nofollow">Qwen2.5-VL-7B-Instruct</a>, specifically the
	<a href="https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct" rel="nofollow">Qwen2.5-VL-7B-Instruct</a> variant.`,name:"text_encoder"},{anchor:"diffusers.HunyuanImageRefinerPipeline.tokenizer",description:"<strong>tokenizer</strong> (<code>Qwen2Tokenizer</code>) — Tokenizer of class [Qwen2Tokenizer].",name:"tokenizer"}]});var u=e(B,4),Z=a(u);n(Z,{name:"__call__",anchor:"diffusers.HunyuanImageRefinerPipeline.__call__",source:"https://github.com/huggingface/diffusers/blob/vr_14192/src/diffusers/pipelines/hunyuan_image/pipeline_hunyuanimage_refiner.py#L436",parameters:[{name:"prompt",val:": str \| list[str] = None"},{name:"negative_prompt",val:": str \| list[str] = None"},{name:"distilled_guidance_scale",val:": float \| None = 3.25"},{name:"image",val:": PipelineImageInput \| None = None"},{name:"height",val:": int \| None = None"},{name:"width",val:": int \| None = None"},{name:"num_inference_steps",val:": int = 4"},{name:"sigmas",val:": list[float] \| None = None"},{name:"num_images_per_prompt",val:": int = 1"},{name:"generator",val:": torch.Generator \| list[torch.Generator] \| None = None"},{name:"latents",val:": torch.Tensor \| None = None"},{name:"prompt_embeds",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_mask",val:": torch.Tensor \| None = None"},{name:"negative_prompt_embeds",val:": torch.Tensor \| None = None"},{name:"negative_prompt_embeds_mask",val:": torch.Tensor \| None = None"},{name:"output_type",val:": str \| None = 'pil'"},{name:"return_dict",val:": bool = True"},{name:"attention_kwargs",val:": dict[str, Any] \| None = None"},{name:"callback_on_step_end",val:": Callable[[int, int], None] \| None = None"},{name:"callback_on_step_end_tensor_inputs",val:": list[str] = ['latents']"}],parametersDescription:[{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>list[str]</code>, <em>optional</em>) —
	The prompt or prompts to guide the image generation. If not defined, one has to pass <code>prompt_embeds</code>.
	instead.`,name:"prompt"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.negative_prompt",description:`<strong>negative_prompt</strong> (<code>str</code> or <code>list[str]</code>, <em>optional</em>) —
	The prompt or prompts not to guide the image generation. If not defined, will use an empty negative
	prompt. Ignored when not using guidance.`,name:"negative_prompt"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.distilled_guidance_scale",description:`<strong>distilled_guidance_scale</strong> (<code>float</code>, <em>optional</em>, defaults to None) —
	A guidance scale value for guidance distilled models. Unlike the traditional classifier-free guidance
	where the guidance scale is applied during inference through noise prediction rescaling, guidance
	distilled models take the guidance scale directly as an input parameter during forward pass. Guidance
	is enabled by setting <code>distilled_guidance_scale > 1</code>. Higher guidance scale encourages to generate
	images that are closely linked to the text <code>prompt</code>, usually at the expense of lower image quality. For
	guidance distilled models, this parameter is required. For non-distilled models, this parameter will be
	ignored.`,name:"distilled_guidance_scale"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.image",description:`<strong>image</strong> (<code>PipelineImageInput</code>, <em>optional</em>) —
	The input image to be refined.`,name:"image"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.num_images_per_prompt",description:"<strong>num_images_per_prompt</strong> (<code>int</code>, <em>optional</em>, defaults to 1) —",name:"num_images_per_prompt"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.height",description:`<strong>height</strong> (<code>int</code>, <em>optional</em>, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
	The height in pixels of the generated image. This is set to 1024 by default for the best results.`,name:"height"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.width",description:`<strong>width</strong> (<code>int</code>, <em>optional</em>, defaults to self.unet.config.sample_size * self.vae_scale_factor) —
	The width in pixels of the generated image. This is set to 1024 by default for the best results.`,name:"width"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.num_inference_steps",description:`<strong>num_inference_steps</strong> (<code>int</code>, <em>optional</em>, defaults to 50) —
	The number of denoising steps. More denoising steps usually lead to a higher quality image at the
	expense of slower inference.`,name:"num_inference_steps"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.sigmas",description:`<strong>sigmas</strong> (<code>list[float]</code>, <em>optional</em>) —
	Custom sigmas to use for the denoising process with schedulers which support a <code>sigmas</code> argument in
	their <code>set_timesteps</code> method. If not defined, the default behavior when <code>num_inference_steps</code> is passed
	will be used.`,name:"sigmas"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.num_images_per_prompt",description:`<strong>num_images_per_prompt</strong> (<code>int</code>, <em>optional</em>, defaults to 1) —
	The number of images to generate per prompt.`,name:"num_images_per_prompt"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.generator",description:`<strong>generator</strong> (<code>torch.Generator</code> or <code>list[torch.Generator]</code>, <em>optional</em>) —
	One or a list of <a href="https://pytorch.org/docs/stable/generated/torch.Generator.html" rel="nofollow">torch generator(s)</a>
	to make generation deterministic.`,name:"generator"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.latents",description:`<strong>latents</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
	generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
	tensor will be generated by sampling using the supplied random <code>generator</code>.`,name:"latents"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.prompt_embeds",description:`<strong>prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings. Can be used to easily tweak text inputs, <em>e.g.</em> prompt weighting. If not
	provided, text embeddings will be generated from <code>prompt</code> input argument.`,name:"prompt_embeds"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.prompt_embeds_mask",description:`<strong>prompt_embeds_mask</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Attention mask for <code>prompt_embeds</code>.`,name:"prompt_embeds_mask"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.negative_prompt_embeds",description:`<strong>negative_prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated negative text embeddings. Can be used to easily tweak text inputs, <em>e.g.</em> prompt
	weighting. If not provided, negative_prompt_embeds will be generated from <code>negative_prompt</code> input
	argument.`,name:"negative_prompt_embeds"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.negative_prompt_embeds_mask",description:`<strong>negative_prompt_embeds_mask</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Attention mask for <code>negative_prompt_embeds</code>.`,name:"negative_prompt_embeds_mask"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.output_type",description:`<strong>output_type</strong> (<code>str</code>, <em>optional</em>, defaults to <code>"pil"</code>) —
	The output format of the generate image. Choose between
	<a href="https://pillow.readthedocs.io/en/stable/" rel="nofollow">PIL</a>: <code>PIL.Image.Image</code> or <code>np.array</code>.`,name:"output_type"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.return_dict",description:`<strong>return_dict</strong> (<code>bool</code>, <em>optional</em>, defaults to <code>True</code>) —
	Whether or not to return a <code>~pipelines.qwenimage.QwenImagePipelineOutput</code> instead of a plain tuple.`,name:"return_dict"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.attention_kwargs",description:`<strong>attention_kwargs</strong> (<code>dict</code>, <em>optional</em>) —
	A kwargs dictionary that if specified is passed along to the <code>AttentionProcessor</code> as defined under
	<code>self.processor</code> in
	<a href="https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py" rel="nofollow">diffusers.models.attention_processor</a>.`,name:"attention_kwargs"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.callback_on_step_end",description:`<strong>callback_on_step_end</strong> (<code>Callable</code>, <em>optional</em>) —
	A function that calls at the end of each denoising steps during the inference. The function is called
	with the following arguments: <code>callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)</code>. <code>callback_kwargs</code> will include a list of all tensors as specified by
	<code>callback_on_step_end_tensor_inputs</code>.`,name:"callback_on_step_end"},{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.callback_on_step_end_tensor_inputs",description:`<strong>callback_on_step_end_tensor_inputs</strong> (<code>List</code>, <em>optional</em>) —
	The list of tensor inputs for the <code>callback_on_step_end</code> function. The tensors specified in the list
	will be passed as <code>callback_kwargs</code> argument. You will only be able to include variables listed in the
	<code>._callback_tensor_inputs</code> attribute of your pipeline class.`,name:"callback_on_step_end_tensor_inputs"}],returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>~pipelines.hunyuan_image.HunyuanImagePipelineOutput</code> if <code>return_dict</code> is True, otherwise a <code>tuple</code>. When
	returning a tuple, the first element is a list with the generated images.</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>~pipelines.hunyuan_image.HunyuanImagePipelineOutput</code> or <code>tuple</code></p>
	`});var Q=e(Z,4);N(Q,{anchor:"diffusers.HunyuanImageRefinerPipeline.__call__.example",children:(o,l)=>{var r=R(),_=e(h(r),2);i(_,{code:"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",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">import</span> torch
	<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> HunyuanImageRefinerPipeline

	<span class="hljs-meta">>>> </span>pipe = HunyuanImageRefinerPipeline.from_pretrained(
	<span class="hljs-meta">... </span> <span class="hljs-string">"hunyuanvideo-community/HunyuanImage-2.1-Refiner-Diffusers"</span>, torch_dtype=torch.bfloat16
	<span class="hljs-meta">... </span>)
	<span class="hljs-meta">>>> </span>pipe.to(<span class="hljs-string">"cuda"</span>)
	<span class="hljs-meta">>>> </span>prompt = <span class="hljs-string">"A cat holding a sign that says hello world"</span>
	<span class="hljs-meta">>>> </span>image = load_image(<span class="hljs-string">"path/to/image.png"</span>)
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># Depending on the variant being used, the pipeline call will slightly vary.</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># Refer to the pipeline documentation for more details.</span>
	<span class="hljs-meta">>>> </span>image = pipe(prompt, image=image, num_inference_steps=<span class="hljs-number">4</span>).images[<span class="hljs-number">0</span>]
	<span class="hljs-meta">>>> </span>image.save(<span class="hljs-string">"hunyuanimage.png"</span>)`,lang:"py",wrap:!1}),d(o,r)},$$slots:{default:!0}}),t(u);var P=e(u,2),X=a(P);n(X,{name:"encode_prompt",anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt",source:"https://github.com/huggingface/diffusers/blob/vr_14192/src/diffusers/pipelines/hunyuan_image/pipeline_hunyuanimage_refiner.py#L225",parameters:[{name:"prompt",val:": str \| list[str] \| None = None"},{name:"device",val:": torch.device \| None = None"},{name:"batch_size",val:": int = 1"},{name:"num_images_per_prompt",val:": int = 1"},{name:"prompt_embeds",val:": torch.Tensor \| None = None"},{name:"prompt_embeds_mask",val:": torch.Tensor \| None = None"}],parametersDescription:[{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>list[str]</code>, <em>optional</em>) —
	prompt to be encoded`,name:"prompt"},{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.device",description:`<strong>device</strong> — (<code>torch.device</code>):
	torch device`,name:"device"},{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.batch_size",description:`<strong>batch_size</strong> (<code>int</code>) —
	batch size of prompts, defaults to 1`,name:"batch_size"},{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.num_images_per_prompt",description:`<strong>num_images_per_prompt</strong> (<code>int</code>) —
	number of images that should be generated per prompt`,name:"num_images_per_prompt"},{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.prompt_embeds",description:`<strong>prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings. If not provided, text embeddings will be generated from <code>prompt</code> input
	argument.`,name:"prompt_embeds"},{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.prompt_embeds_mask",description:`<strong>prompt_embeds_mask</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text mask. If not provided, text mask will be generated from <code>prompt</code> input argument.`,name:"prompt_embeds_mask"},{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.prompt_embeds_2",description:`<strong>prompt_embeds_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated glyph text embeddings from ByT5. If not provided, will be generated from <code>prompt</code> input
	argument using self.tokenizer_2 and self.text_encoder_2.`,name:"prompt_embeds_2"},{anchor:"diffusers.HunyuanImageRefinerPipeline.encode_prompt.prompt_embeds_mask_2",description:`<strong>prompt_embeds_mask_2</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated glyph text mask from ByT5. If not provided, will be generated from <code>prompt</code> input
	argument using self.tokenizer_2 and self.text_encoder_2.`,name:"prompt_embeds_mask_2"}]}),t(P),t(m);var G=e(m,2);s(G,{title:"HunyuanImagePipelineOutput",local:"diffusers.pipelines.hunyuan_image.pipeline_output.HunyuanImagePipelineOutput",headingTag:"h2"});var g=e(G,2),z=a(g);n(z,{name:"class diffusers.pipelines.hunyuan_image.pipeline_output.HunyuanImagePipelineOutput",anchor:"diffusers.pipelines.hunyuan_image.pipeline_output.HunyuanImagePipelineOutput",source:"https://github.com/huggingface/diffusers/blob/vr_14192/src/diffusers/pipelines/hunyuan_image/pipeline_output.py#L10",parameters:[{name:"images",val:": list"}],parametersDescription:[{anchor:"diffusers.pipelines.hunyuan_image.pipeline_output.HunyuanImagePipelineOutput.images",description:`<strong>images</strong> (<code>list[PIL.Image.Image]</code> or <code>np.ndarray</code>) —
	List of denoised PIL images of length <code>batch_size</code> or numpy array of shape <code>(batch_size, height, width, num_channels)</code>. PIL images or numpy array present the denoised images of the diffusion pipeline.`,name:"images"}]}),W(2),t(g);var Y=e(g,2);D(Y,{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/pipelines/hunyuanimage21.md"}),W(2),d(C,y),O()}export{ie as component};

Xet Storage Details

Size:: 53.4 kB
Xet hash:: f4b3b2fd0891ed80c85358a6a234e74ff5f663892c5aef00d37c4db785129c90

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