Buckets:

hf-doc-build
/

doc-dev

hf-doc-build/doc-dev / diffusers /pr_11636 /en /_app /immutable /nodes /152.da6701cd.js

rtrm's picture

29 days ago

102 kB

	import{s as Hs,o as As,n as Ss}from"../chunks/scheduler.53228c21.js";import{S as zs,i as qs,e as o,s,c as d,h as Ds,a as r,d as t,b as a,f as U,g as c,j as p,k,w as R,l as i,m as l,n as m,t as u,o as y,p as g}from"../chunks/index.100fac89.js";import{C as Os}from"../chunks/CopyLLMTxtMenu.8a16ebe2.js";import{D as x}from"../chunks/Docstring.07ca7ce7.js";import{C as q}from"../chunks/CodeBlock.d30a6509.js";import{E as Ys}from"../chunks/ExampleCodeBlock.672157f9.js";import{H as C,E as ea}from"../chunks/MermaidChart.svelte_svelte_type_style_lang.83a5c0e1.js";function na(Se){let f,G="Examples:",J,T,v;return T=new q({props:{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> Kandinsky5T2VPipeline
	<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> export_to_video

	<span class="hljs-meta">>>> </span><span class="hljs-comment"># Available models:</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Pro-sft-5s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-sft-5s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-nocfg-5s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-distilled16steps-5s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-pretrain-5s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-sft-10s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-nocfg-10s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-distilled16steps-10s-Diffusers</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-T2V-Lite-pretrain-10s-Diffusers</span>

	<span class="hljs-meta">>>> </span>model_id = <span class="hljs-string">"kandinskylab/Kandinsky-5.0-T2V-Lite-sft-5s-Diffusers"</span>
	<span class="hljs-meta">>>> </span>pipe = Kandinsky5T2VPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)
	<span class="hljs-meta">>>> </span>pipe = pipe.to(<span class="hljs-string">"cuda"</span>)

	<span class="hljs-meta">>>> </span>prompt = <span class="hljs-string">"A cat and a dog baking a cake together in a kitchen."</span>
	<span class="hljs-meta">>>> </span>negative_prompt = <span class="hljs-string">"Static, 2D cartoon, cartoon, 2d animation, paintings, images, worst quality, low quality, ugly, deformed, walking backwards"</span>

	<span class="hljs-meta">>>> </span>output = pipe(
	<span class="hljs-meta">... </span> prompt=prompt,
	<span class="hljs-meta">... </span> negative_prompt=negative_prompt,
	<span class="hljs-meta">... </span> height=<span class="hljs-number">512</span>,
	<span class="hljs-meta">... </span> width=<span class="hljs-number">768</span>,
	<span class="hljs-meta">... </span> num_frames=<span class="hljs-number">121</span>,
	<span class="hljs-meta">... </span> num_inference_steps=<span class="hljs-number">50</span>,
	<span class="hljs-meta">... </span> guidance_scale=<span class="hljs-number">5.0</span>,
	<span class="hljs-meta">... </span>).frames[<span class="hljs-number">0</span>]

	<span class="hljs-meta">>>> </span>export_to_video(output, <span class="hljs-string">"output.mp4"</span>, fps=<span class="hljs-number">24</span>, quality=<span class="hljs-number">9</span>)`,wrap:!1}}),{c(){f=o("p"),f.textContent=G,J=s(),d(T.$$.fragment)},l(h){f=r(h,"P",{"data-svelte-h":!0}),p(f)!=="svelte-kvfsh7"&&(f.textContent=G),J=a(h),c(T.$$.fragment,h)},m(h,j){l(h,f,j),l(h,J,j),m(T,h,j),v=!0},p:Ss,i(h){v\|\|(u(T.$$.fragment,h),v=!0)},o(h){y(T.$$.fragment,h),v=!1},d(h){h&&(t(f),t(J)),g(T,h)}}}function ta(Se){let f,G="Examples:",J,T,v;return T=new q({props:{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> Kandinsky5I2VPipeline
	<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> export_to_video, load_image

	<span class="hljs-meta">>>> </span><span class="hljs-comment"># Available models:</span>
	<span class="hljs-meta">>>> </span><span class="hljs-comment"># kandinskylab/Kandinsky-5.0-I2V-Pro-sft-5s-Diffusers</span>

	<span class="hljs-meta">>>> </span>model_id = <span class="hljs-string">"kandinskylab/Kandinsky-5.0-I2V-Pro-sft-5s-Diffusers"</span>
	<span class="hljs-meta">>>> </span>pipe = Kandinsky5I2VPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)
	<span class="hljs-meta">>>> </span>pipe = pipe.to(<span class="hljs-string">"cuda"</span>)

	<span class="hljs-meta">>>> </span>image = load_image(
	<span class="hljs-meta">... </span> <span class="hljs-string">"https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg"</span>
	<span class="hljs-meta">... </span>)
	<span class="hljs-meta">>>> </span>prompt = <span class="hljs-string">"An astronaut floating in space with Earth in the background, cinematic shot"</span>
	<span class="hljs-meta">>>> </span>negative_prompt = <span class="hljs-string">"Static, 2D cartoon, cartoon, 2d animation, paintings, images, worst quality, low quality, ugly, deformed, walking backwards"</span>

	<span class="hljs-meta">>>> </span>output = pipe(
	<span class="hljs-meta">... </span> image=image,
	<span class="hljs-meta">... </span> prompt=prompt,
	<span class="hljs-meta">... </span> negative_prompt=negative_prompt,
	<span class="hljs-meta">... </span> height=<span class="hljs-number">512</span>,
	<span class="hljs-meta">... </span> width=<span class="hljs-number">768</span>,
	<span class="hljs-meta">... </span> num_frames=<span class="hljs-number">121</span>,
	<span class="hljs-meta">... </span> num_inference_steps=<span class="hljs-number">50</span>,
	<span class="hljs-meta">... </span> guidance_scale=<span class="hljs-number">5.0</span>,
	<span class="hljs-meta">... </span>).frames[<span class="hljs-number">0</span>]

	<span class="hljs-meta">>>> </span>export_to_video(output, <span class="hljs-string">"output.mp4"</span>, fps=<span class="hljs-number">24</span>, quality=<span class="hljs-number">9</span>)`,wrap:!1}}),{c(){f=o("p"),f.textContent=G,J=s(),d(T.$$.fragment)},l(h){f=r(h,"P",{"data-svelte-h":!0}),p(f)!=="svelte-kvfsh7"&&(f.textContent=G),J=a(h),c(T.$$.fragment,h)},m(h,j){l(h,f,j),l(h,J,j),m(T,h,j),v=!0},p:Ss,i(h){v\|\|(u(T.$$.fragment,h),v=!0)},o(h){y(T.$$.fragment,h),v=!1},d(h){h&&(t(f),t(J)),g(T,h)}}}function sa(Se){let f,G,J,T,v,h,j,xn,D,as='<a href="https://arxiv.org/abs/2511.14993" rel="nofollow">Kandinsky 5.0</a> is a family of diffusion models for Video & Image generation.',In,O,is="Kandinsky 5.0 Lite line-up of lightweight video generation models (2B parameters) that ranks #1 among open-source models in its class. It outperforms larger models and offers the best understanding of Russian concepts in the open-source ecosystem.",Vn,ee,ls="Kandinsky 5.0 Pro line-up of large high quality video generation models (19B parameters). It offers high qualty generation in HD and more generation formats like I2V.",Bn,ne,os="The model introduces several key innovations:",Gn,te,rs="<li><strong>Latent diffusion pipeline</strong> with <strong>Flow Matching</strong> for improved training stability</li> <li><strong>Diffusion Transformer (DiT)</strong> as the main generative backbone with cross-attention to text embeddings</li> <li>Dual text encoding using <strong>Qwen2.5-VL</strong> and <strong>CLIP</strong> for comprehensive text understanding</li> <li><strong>HunyuanVideo 3D VAE</strong> for efficient video encoding and decoding</li> <li><strong>Sparse attention mechanisms</strong> (NABLA) for efficient long-sequence processing</li>",Zn,se,ps='The original codebase can be found at <a href="https://github.com/kandinskylab/Kandinsky-5" rel="nofollow">kandinskylab/Kandinsky-5</a>.',Wn,Y,ds='<p>Check out the <a href="https://huggingface.co/kandinskylab" rel="nofollow">Kandinsky Lab</a> organization on the Hub for the official model checkpoints for text-to-video generation, including pretrained, SFT, no-CFG, and distilled variants.</p>',$n,ae,Pn,ie,cs="Kandinsky 5.0 T2V Pro:",Qn,le,ms="<thead><tr><th>model_id</th> <th>Description</th> <th>Use Cases</th></tr></thead> <tbody><tr><td><strong>kandinskylab/Kandinsky-5.0-T2V-Pro-sft-5s-Diffusers</strong></td> <td>5 second Text-to-Video Pro model</td> <td>High-quality text-to-video generation</td></tr> <tr><td><strong>kandinskylab/Kandinsky-5.0-I2V-Pro-sft-5s-Diffusers</strong></td> <td>5 second Image-to-Video Pro model</td> <td>High-quality image-to-video generation</td></tr></tbody>",Kn,oe,us=`Kandinsky 5.0 T2V Lite:
	\| model_id \| Description \| Use Cases \|
	\|------------\|-------------\|-----------\|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-sft-5s-Diffusers</strong> \| 5 second Supervised Fine-Tuned model \| Highest generation quality \|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-sft-10s-Diffusers</strong> \| 10 second Supervised Fine-Tuned model \| Highest generation quality \|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-nocfg-5s-Diffusers</strong> \| 5 second Classifier-Free Guidance distilled \| 2× faster inference \|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-nocfg-10s-Diffusers</strong> \| 10 second Classifier-Free Guidance distilled \| 2× faster inference \|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-distilled16steps-5s-Diffusers</strong> \| 5 second Diffusion distilled to 16 steps \| 6× faster inference, minimal quality loss \|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-distilled16steps-10s-Diffusers</strong> \| 10 second Diffusion distilled to 16 steps \| 6× faster inference, minimal quality loss \|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-pretrain-5s-Diffusers</strong> \| 5 second Base pretrained model \| Research and fine-tuning \|
	\| <strong>kandinskylab/Kandinsky-5.0-T2V-Lite-pretrain-10s-Diffusers</strong> \| 10 second Base pretrained model \| Research and fine-tuning \|`,En,re,Nn,pe,Xn,de,Fn,ce,ys="<strong>⚠️ Warning!</strong> all Pro models should be infered with pipeline.enable_model_cpu_offload()",Ln,me,Rn,ue,Yn,ye,Sn,ge,Hn,he,gs="<strong>⚠️ Warning!</strong> all 10 second models should be used with Flex attention and max-autotune-no-cudagraphs compilation:",An,fe,zn,_e,qn,Me,hs="<strong>⚠️ Warning!</strong> all nocfg and diffusion distilled models should be infered wothout CFG (<code>guidance_scale=1.0</code>):",Dn,be,On,Te,et,ke,fs="<strong>⚠️ Warning!</strong> all Pro models should be infered with pipeline.enable_model_cpu_offload()",nt,we,tt,ve,st,I,_s='<tr><td><img width="200" alt="image" src="https://github.com/user-attachments/assets/73e5ff00-2735-40fd-8f01-767de9181918"/></td> <td><img width="200" alt="image" src="https://github.com/user-attachments/assets/f449a9e7-74b7-481d-82da-02723e396acd"/></td> </tr><tr><td>Comparison with Veo 3</td> <td>Comparison with Veo 3 fast</td> </tr><tr><td><img width="200" alt="image" src="https://github.com/user-attachments/assets/a6902fb6-b5e8-4093-adad-aa4caab79c6d"/></td> <td><img width="200" alt="image" src="https://github.com/user-attachments/assets/09986015-3d07-4de8-b942-c145039b9b2d"/></td> </tr><tr><td>Comparison with Wan 2.2 A14B Text-to-Video mode</td> <td>Comparison with Wan 2.2 A14B Image-to-Video mode</td></tr>',at,Ue,it,Je,Ms='The evaluation is based on the expanded prompts from the <a href="https://github.com/facebookresearch/MovieGenBench" rel="nofollow">Movie Gen benchmark</a>, which are available in the expanded_prompt column of the benchmark/moviegen_bench.csv file.',lt,V,bs='<tr><td><img src="https://github.com/kandinskylab/kandinsky-5/raw/main/assets/sbs/kandinsky_5_video_lite_vs_sora.jpg" width="400"/></td> <td><img src="https://github.com/kandinskylab/kandinsky-5/raw/main/assets/sbs/kandinsky_5_video_lite_vs_wan_2.1_14B.jpg" width="400"/></td> </tr><tr><td><img src="https://github.com/kandinskylab/kandinsky-5/raw/main/assets/sbs/kandinsky_5_video_lite_vs_wan_2.2_5B.jpg" width="400"/></td> <td><img src="https://github.com/kandinskylab/kandinsky-5/raw/main/assets/sbs/kandinsky_5_video_lite_vs_wan_2.2_A14B.jpg" width="400"/></td> </tr><tr><td><img src="https://github.com/kandinskylab/kandinsky-5/raw/main/assets/sbs/kandinsky_5_video_lite_vs_wan_2.1_1.3B.jpg" width="400"/></td></tr>',ot,je,rt,B,Ts='<tr><td><img src="https://github.com/kandinskylab/kandinsky-5/raw/main/assets/sbs/kandinsky_5_video_lite_5s_vs_kandinsky_5_video_lite_distill_5s.jpg" width="400"/></td> <td><img src="https://github.com/kandinskylab/kandinsky-5/raw/main/assets/sbs/kandinsky_5_video_lite_10s_vs_kandinsky_5_video_lite_distill_10s.jpg" width="400"/></td></tr>',pt,Ce,dt,_,xe,bt,He,ks="Pipeline for text-to-video generation using Kandinsky 5.0.",Tt,Ae,ws=`This model inherits from <a href="/docs/diffusers/pr_11636/en/api/pipelines/overview#diffusers.DiffusionPipeline">DiffusionPipeline</a>. Check the superclass documentation for the generic methods
	implemented for all pipelines (downloading, saving, running on a particular device, etc.).`,kt,Z,Ie,wt,ze,vs="The call function to the pipeline for generation.",vt,S,Ut,H,Ve,Jt,qe,Us="Validate input parameters for the pipeline.",jt,W,Be,Ct,De,Js="Encodes a single prompt (positive or negative) into text encoder hidden states.",xt,Oe,js=`This method combines embeddings from both Qwen2.5-VL and CLIP text encoders to create comprehensive text
	representations for video generation.`,It,$,Ge,Vt,en,Cs="Create a sparse temporal attention (STA) mask for efficient video generation.",Bt,nn,xs=`This method generates a mask that limits attention to nearby frames and spatial positions, reducing
	computational complexity for video generation.`,Gt,P,Ze,Zt,tn,Is="Generate sparse attention parameters for the transformer based on sample dimensions.",Wt,sn,Vs=`This method computes the sparse attention configuration needed for efficient video processing in the
	transformer model.`,$t,Q,We,Pt,an,Bs="Prepare initial latent variables for video generation.",Qt,ln,Gs="This method creates random noise latents or uses provided latents as starting point for the denoising process.",ct,$e,mt,M,Pe,Kt,on,Zs="Pipeline for image-to-video generation using Kandinsky 5.0.",Et,rn,Ws=`This model inherits from <a href="/docs/diffusers/pr_11636/en/api/pipelines/overview#diffusers.DiffusionPipeline">DiffusionPipeline</a>. Check the superclass documentation for the generic methods
	implemented for all pipelines (downloading, saving, running on a particular device, etc.).`,Nt,K,Qe,Xt,pn,$s="The call function to the pipeline for image-to-video generation.",Ft,A,Lt,z,Ke,Rt,dn,Ps="Validate input parameters for the pipeline.",Yt,E,Ee,St,cn,Qs="Encodes a single prompt (positive or negative) into text encoder hidden states.",Ht,mn,Ks=`This method combines embeddings from both Qwen2.5-VL and CLIP text encoders to create comprehensive text
	representations for video generation.`,At,N,Ne,zt,un,Es="Create a sparse temporal attention (STA) mask for efficient video generation.",qt,yn,Ns=`This method generates a mask that limits attention to nearby frames and spatial positions, reducing
	computational complexity for video generation.`,Dt,X,Xe,Ot,gn,Xs="Generate sparse attention parameters for the transformer based on sample dimensions.",es,hn,Fs=`This method computes the sparse attention configuration needed for efficient video processing in the
	transformer model.`,ns,F,Fe,ts,fn,Ls="Prepare initial latent variables for image-to-video generation.",ss,_n,Rs=`This method creates random noise latents for all frames except the first frame, which is replaced with the
	encoded input image.`,ut,Le,yt,Re,gt,Ye,ht,Cn,ft;return v=new Os({props:{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"}}),j=new C({props:{title:"Kandinsky 5.0 Video",local:"kandinsky-50-video",headingTag:"h1"}}),ae=new C({props:{title:"Available Models",local:"available-models",headingTag:"h2"}}),re=new C({props:{title:"Usage Examples",local:"usage-examples",headingTag:"h2"}}),pe=new C({props:{title:"Basic Text-to-Video Generation",local:"basic-text-to-video-generation",headingTag:"h3"}}),de=new C({props:{title:"Pro",local:"pro",headingTag:"h4"}}),me=new q({props:{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> Kandinsky5T2VPipeline
	<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> export_to_video

	<span class="hljs-comment"># Load the pipeline</span>
	model_id = <span class="hljs-string">"kandinskylab/Kandinsky-5.0-T2V-Pro-sft-5s-Diffusers"</span>
	pipe = Kandinsky5T2VPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)

	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)
	pipeline.transformer.set_attention_backend(<span class="hljs-string">"flex"</span>) <span class="hljs-comment"># <--- Set attention bakend to Flex</span>
	pipeline.enable_model_cpu_offload() <span class="hljs-comment"># <--- Enable cpu offloading for single GPU inference</span>
	pipeline.transformer.<span class="hljs-built_in">compile</span>(mode=<span class="hljs-string">"max-autotune-no-cudagraphs"</span>, dynamic=<span class="hljs-literal">True</span>) <span class="hljs-comment"># <--- Compile with max-autotune-no-cudagraphs</span>

	<span class="hljs-comment"># Generate video</span>
	prompt = <span class="hljs-string">"A cat and a dog baking a cake together in a kitchen."</span>
	negative_prompt = <span class="hljs-string">"Static, 2D cartoon, cartoon, 2d animation, paintings, images, worst quality, low quality, ugly, deformed, walking backwards"</span>

	output = pipe(
	prompt=prompt,
	negative_prompt=negative_prompt,
	height=<span class="hljs-number">768</span>,
	width=<span class="hljs-number">1024</span>,
	num_frames=<span class="hljs-number">121</span>, <span class="hljs-comment"># ~5 seconds at 24fps</span>
	num_inference_steps=<span class="hljs-number">50</span>,
	guidance_scale=<span class="hljs-number">5.0</span>,
	).frames[<span class="hljs-number">0</span>]

	export_to_video(output, <span class="hljs-string">"output.mp4"</span>, fps=<span class="hljs-number">24</span>, quality=<span class="hljs-number">9</span>)`,wrap:!1}}),ue=new C({props:{title:"Lite",local:"lite",headingTag:"h4"}}),ye=new q({props:{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> Kandinsky5T2VPipeline
	<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> export_to_video

	<span class="hljs-comment"># Load the pipeline</span>
	model_id = <span class="hljs-string">"kandinskylab/Kandinsky-5.0-T2V-Lite-sft-5s-Diffusers"</span>
	pipe = Kandinsky5T2VPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)
	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)

	<span class="hljs-comment"># Generate video</span>
	prompt = <span class="hljs-string">"A cat and a dog baking a cake together in a kitchen."</span>
	negative_prompt = <span class="hljs-string">"Static, 2D cartoon, cartoon, 2d animation, paintings, images, worst quality, low quality, ugly, deformed, walking backwards"</span>

	output = pipe(
	prompt=prompt,
	negative_prompt=negative_prompt,
	height=<span class="hljs-number">512</span>,
	width=<span class="hljs-number">768</span>,
	num_frames=<span class="hljs-number">121</span>, <span class="hljs-comment"># ~5 seconds at 24fps</span>
	num_inference_steps=<span class="hljs-number">50</span>,
	guidance_scale=<span class="hljs-number">5.0</span>,
	).frames[<span class="hljs-number">0</span>]

	export_to_video(output, <span class="hljs-string">"output.mp4"</span>, fps=<span class="hljs-number">24</span>, quality=<span class="hljs-number">9</span>)`,wrap:!1}}),ge=new C({props:{title:"10 second Models",local:"10-second-models",headingTag:"h3"}}),fe=new q({props:{code:"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",highlighted:`pipe = Kandinsky5T2VPipeline.from_pretrained(
	<span class="hljs-string">"kandinskylab/Kandinsky-5.0-T2V-Lite-sft-10s-Diffusers"</span>,
	torch_dtype=torch.bfloat16
	)
	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)

	pipe.transformer.set_attention_backend(
	<span class="hljs-string">"flex"</span>
	) <span class="hljs-comment"># <--- Set attention bakend to Flex</span>
	pipe.transformer.<span class="hljs-built_in">compile</span>(
	mode=<span class="hljs-string">"max-autotune-no-cudagraphs"</span>,
	dynamic=<span class="hljs-literal">True</span>
	) <span class="hljs-comment"># <--- Compile with max-autotune-no-cudagraphs</span>

	prompt = <span class="hljs-string">"A cat and a dog baking a cake together in a kitchen."</span>
	negative_prompt = <span class="hljs-string">"Static, 2D cartoon, cartoon, 2d animation, paintings, images, worst quality, low quality, ugly, deformed, walking backwards"</span>

	output = pipe(
	prompt=prompt,
	negative_prompt=negative_prompt,
	height=<span class="hljs-number">512</span>,
	width=<span class="hljs-number">768</span>,
	num_frames=<span class="hljs-number">241</span>,
	num_inference_steps=<span class="hljs-number">50</span>,
	guidance_scale=<span class="hljs-number">5.0</span>,
	).frames[<span class="hljs-number">0</span>]

	export_to_video(output, <span class="hljs-string">"output.mp4"</span>, fps=<span class="hljs-number">24</span>, quality=<span class="hljs-number">9</span>)`,wrap:!1}}),_e=new C({props:{title:"Diffusion Distilled model",local:"diffusion-distilled-model",headingTag:"h3"}}),be=new q({props:{code:"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",highlighted:`model_id = <span class="hljs-string">"kandinskylab/Kandinsky-5.0-T2V-Lite-distilled16steps-5s-Diffusers"</span>
	pipe = Kandinsky5T2VPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)
	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)

	output = pipe(
	prompt=<span class="hljs-string">"A beautiful sunset over mountains"</span>,
	num_inference_steps=<span class="hljs-number">16</span>, <span class="hljs-comment"># <--- Model is distilled in 16 steps</span>
	guidance_scale=<span class="hljs-number">1.0</span>, <span class="hljs-comment"># <--- no CFG</span>
	).frames[<span class="hljs-number">0</span>]

	export_to_video(output, <span class="hljs-string">"output.mp4"</span>, fps=<span class="hljs-number">24</span>, quality=<span class="hljs-number">9</span>)`,wrap:!1}}),Te=new C({props:{title:"Basic Image-to-Video Generation",local:"basic-image-to-video-generation",headingTag:"h3"}}),we=new q({props:{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> Kandinsky5T2VPipeline
	<span class="hljs-keyword">from</span> diffusers.utils <span class="hljs-keyword">import</span> export_to_video

	<span class="hljs-comment"># Load the pipeline</span>
	model_id = <span class="hljs-string">"kandinskylab/Kandinsky-5.0-I2V-Pro-sft-5s-Diffusers"</span>
	pipe = Kandinsky5T2VPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)

	pipe = pipe.to(<span class="hljs-string">"cuda"</span>)
	pipeline.transformer.set_attention_backend(<span class="hljs-string">"flex"</span>) <span class="hljs-comment"># <--- Set attention bakend to Flex</span>
	pipeline.enable_model_cpu_offload() <span class="hljs-comment"># <--- Enable cpu offloading for single GPU inference</span>
	pipeline.transformer.<span class="hljs-built_in">compile</span>(mode=<span class="hljs-string">"max-autotune-no-cudagraphs"</span>, dynamic=<span class="hljs-literal">True</span>) <span class="hljs-comment"># <--- Compile with max-autotune-no-cudagraphs</span>

	<span class="hljs-comment"># Generate video</span>
	image = load_image(
	<span class="hljs-string">"https://huggingface.co/kandinsky-community/kandinsky-3/resolve/main/assets/title.jpg?download=true"</span>
	)
	height = <span class="hljs-number">896</span>
	width = <span class="hljs-number">896</span>
	image = image.resize((width, height))

	prompt = <span class="hljs-string">"An funny furry creture smiles happily and holds a sign that says 'Kandinsky'"</span>
	negative_prompt = <span class="hljs-string">""</span>

	output = pipe(
	prompt=prompt,
	negative_prompt=negative_prompt,
	height=height,
	width=width,
	num_frames=<span class="hljs-number">121</span>, <span class="hljs-comment"># ~5 seconds at 24fps</span>
	num_inference_steps=<span class="hljs-number">50</span>,
	guidance_scale=<span class="hljs-number">5.0</span>,
	).frames[<span class="hljs-number">0</span>]

	export_to_video(output, <span class="hljs-string">"output.mp4"</span>, fps=<span class="hljs-number">24</span>, quality=<span class="hljs-number">9</span>)`,wrap:!1}}),ve=new C({props:{title:"Kandinsky 5.0 Pro Side-by-Side evaluation",local:"kandinsky-50-pro-side-by-side-evaluation",headingTag:"h2"}}),Ue=new C({props:{title:"Kandinsky 5.0 Lite Side-by-Side evaluation",local:"kandinsky-50-lite-side-by-side-evaluation",headingTag:"h2"}}),je=new C({props:{title:"Kandinsky 5.0 Lite Distill Side-by-Side evaluation",local:"kandinsky-50-lite-distill-side-by-side-evaluation",headingTag:"h2"}}),Ce=new C({props:{title:"Kandinsky5T2VPipeline",local:"diffusers.Kandinsky5T2VPipeline",headingTag:"h2"}}),xe=new x({props:{name:"class diffusers.Kandinsky5T2VPipeline",anchor:"diffusers.Kandinsky5T2VPipeline",parameters:[{name:"transformer",val:": Kandinsky5Transformer3DModel"},{name:"vae",val:": AutoencoderKLHunyuanVideo"},{name:"text_encoder",val:": Qwen2_5_VLForConditionalGeneration"},{name:"tokenizer",val:": Qwen2VLProcessor"},{name:"text_encoder_2",val:": CLIPTextModel"},{name:"tokenizer_2",val:": CLIPTokenizer"},{name:"scheduler",val:": FlowMatchEulerDiscreteScheduler"}],parametersDescription:[{anchor:"diffusers.Kandinsky5T2VPipeline.transformer",description:`<strong>transformer</strong> (<code>Kandinsky5Transformer3DModel</code>) —
	Conditional Transformer to denoise the encoded video latents.`,name:"transformer"},{anchor:"diffusers.Kandinsky5T2VPipeline.vae",description:`<strong>vae</strong> (<a href="/docs/diffusers/pr_11636/en/api/models/autoencoder_kl_hunyuan_video#diffusers.AutoencoderKLHunyuanVideo">AutoencoderKLHunyuanVideo</a>) —
	Variational Auto-Encoder Model <a href="https://huggingface.co/hunyuanvideo-community/HunyuanVideo" rel="nofollow">hunyuanvideo-community/HunyuanVideo
	(vae)</a> to encode and decode videos to and from
	latent representations.`,name:"vae"},{anchor:"diffusers.Kandinsky5T2VPipeline.text_encoder",description:`<strong>text_encoder</strong> (<code>Qwen2_5_VLForConditionalGeneration</code>) —
	Frozen text-encoder <a href="https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct" rel="nofollow">Qwen2.5-VL</a>.`,name:"text_encoder"},{anchor:"diffusers.Kandinsky5T2VPipeline.tokenizer",description:`<strong>tokenizer</strong> (<code>AutoProcessor</code>) —
	Tokenizer for Qwen2.5-VL.`,name:"tokenizer"},{anchor:"diffusers.Kandinsky5T2VPipeline.text_encoder_2",description:`<strong>text_encoder_2</strong> (<code>CLIPTextModel</code>) —
	Frozen <a href="https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel" rel="nofollow">CLIP</a>,
	specifically the <a href="https://huggingface.co/openai/clip-vit-large-patch14" rel="nofollow">clip-vit-large-patch14</a> variant.`,name:"text_encoder_2"},{anchor:"diffusers.Kandinsky5T2VPipeline.tokenizer_2",description:`<strong>tokenizer_2</strong> (<code>CLIPTokenizer</code>) —
	Tokenizer for CLIP.`,name:"tokenizer_2"},{anchor:"diffusers.Kandinsky5T2VPipeline.scheduler",description:`<strong>scheduler</strong> (<a href="/docs/diffusers/pr_11636/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 video latents.`,name:"scheduler"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky.py#L131"}}),Ie=new x({props:{name:"__call__",anchor:"diffusers.Kandinsky5T2VPipeline.__call__",parameters:[{name:"prompt",val:": typing.Union[str, typing.List[str]] = None"},{name:"negative_prompt",val:": typing.Union[str, typing.List[str], NoneType] = None"},{name:"height",val:": int = 512"},{name:"width",val:": int = 768"},{name:"num_frames",val:": int = 121"},{name:"num_inference_steps",val:": int = 50"},{name:"guidance_scale",val:": float = 5.0"},{name:"num_videos_per_prompt",val:": typing.Optional[int] = 1"},{name:"generator",val:": typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None"},{name:"latents",val:": typing.Optional[torch.Tensor] = None"},{name:"prompt_embeds_qwen",val:": typing.Optional[torch.Tensor] = None"},{name:"prompt_embeds_clip",val:": typing.Optional[torch.Tensor] = None"},{name:"negative_prompt_embeds_qwen",val:": typing.Optional[torch.Tensor] = None"},{name:"negative_prompt_embeds_clip",val:": typing.Optional[torch.Tensor] = None"},{name:"prompt_cu_seqlens",val:": typing.Optional[torch.Tensor] = None"},{name:"negative_prompt_cu_seqlens",val:": typing.Optional[torch.Tensor] = None"},{name:"output_type",val:": typing.Optional[str] = 'pil'"},{name:"return_dict",val:": bool = True"},{name:"callback_on_step_end",val:": typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], diffusers.callbacks.PipelineCallback, diffusers.callbacks.MultiPipelineCallbacks, NoneType] = None"},{name:"callback_on_step_end_tensor_inputs",val:": typing.List[str] = ['latents']"},{name:"max_sequence_length",val:": int = 512"}],parametersDescription:[{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>List[str]</code>, <em>optional</em>) —
	The prompt or prompts to guide the video generation. If not defined, pass <code>prompt_embeds</code> instead.`,name:"prompt"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.negative_prompt",description:`<strong>negative_prompt</strong> (<code>str</code> or <code>List[str]</code>, <em>optional</em>) —
	The prompt or prompts to avoid during video generation. If not defined, pass <code>negative_prompt_embeds</code>
	instead. Ignored when not using guidance (<code>guidance_scale</code> < <code>1</code>).`,name:"negative_prompt"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.height",description:`<strong>height</strong> (<code>int</code>, defaults to <code>512</code>) —
	The height in pixels of the generated video.`,name:"height"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.width",description:`<strong>width</strong> (<code>int</code>, defaults to <code>768</code>) —
	The width in pixels of the generated video.`,name:"width"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.num_frames",description:`<strong>num_frames</strong> (<code>int</code>, defaults to <code>25</code>) —
	The number of frames in the generated video.`,name:"num_frames"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.num_inference_steps",description:`<strong>num_inference_steps</strong> (<code>int</code>, defaults to <code>50</code>) —
	The number of denoising steps.`,name:"num_inference_steps"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.guidance_scale",description:`<strong>guidance_scale</strong> (<code>float</code>, defaults to <code>5.0</code>) —
	Guidance scale as defined in classifier-free guidance.`,name:"guidance_scale"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.num_videos_per_prompt",description:`<strong>num_videos_per_prompt</strong> (<code>int</code>, <em>optional</em>, defaults to 1) —
	The number of videos to generate per prompt.`,name:"num_videos_per_prompt"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.generator",description:`<strong>generator</strong> (<code>torch.Generator</code> or <code>List[torch.Generator]</code>, <em>optional</em>) —
	A torch generator to make generation deterministic.`,name:"generator"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.latents",description:`<strong>latents</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated noisy latents.`,name:"latents"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.prompt_embeds",description:`<strong>prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated text embeddings.`,name:"prompt_embeds"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.negative_prompt_embeds",description:`<strong>negative_prompt_embeds</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated negative text embeddings.`,name:"negative_prompt_embeds"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.output_type",description:`<strong>output_type</strong> (<code>str</code>, <em>optional</em>, defaults to <code>"pil"</code>) —
	The output format of the generated video.`,name:"output_type"},{anchor:"diffusers.Kandinsky5T2VPipeline.__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>KandinskyPipelineOutput</code>.`,name:"return_dict"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.callback_on_step_end",description:`<strong>callback_on_step_end</strong> (<code>Callable</code>, <code>PipelineCallback</code>, <code>MultiPipelineCallbacks</code>, <em>optional</em>) —
	A function that is called at the end of each denoising step.`,name:"callback_on_step_end"},{anchor:"diffusers.Kandinsky5T2VPipeline.__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.`,name:"callback_on_step_end_tensor_inputs"},{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.max_sequence_length",description:`<strong>max_sequence_length</strong> (<code>int</code>, defaults to <code>512</code>) —
	The maximum sequence length for text encoding.`,name:"max_sequence_length"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky.py#L682",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>If <code>return_dict</code> is <code>True</code>, <code>KandinskyPipelineOutput</code> is returned, otherwise a <code>tuple</code> is returned
	where the first element is a list with the generated images.</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>~KandinskyPipelineOutput</code> or <code>tuple</code></p>
	`}}),S=new Ys({props:{anchor:"diffusers.Kandinsky5T2VPipeline.__call__.example",$$slots:{default:[na]},$$scope:{ctx:Se}}}),Ve=new x({props:{name:"check_inputs",anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs",parameters:[{name:"prompt",val:""},{name:"negative_prompt",val:""},{name:"height",val:""},{name:"width",val:""},{name:"prompt_embeds_qwen",val:" = None"},{name:"prompt_embeds_clip",val:" = None"},{name:"negative_prompt_embeds_qwen",val:" = None"},{name:"negative_prompt_embeds_clip",val:" = None"},{name:"prompt_cu_seqlens",val:" = None"},{name:"negative_prompt_cu_seqlens",val:" = None"},{name:"callback_on_step_end_tensor_inputs",val:" = None"},{name:"max_sequence_length",val:" = None"}],parametersDescription:[{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.prompt",description:"<strong>prompt</strong> — Input prompt",name:"prompt"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.negative_prompt",description:"<strong>negative_prompt</strong> — Negative prompt for guidance",name:"negative_prompt"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.height",description:"<strong>height</strong> — Video height",name:"height"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.width",description:"<strong>width</strong> — Video width",name:"width"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.prompt_embeds_qwen",description:"<strong>prompt_embeds_qwen</strong> — Pre-computed Qwen prompt embeddings",name:"prompt_embeds_qwen"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.prompt_embeds_clip",description:"<strong>prompt_embeds_clip</strong> — Pre-computed CLIP prompt embeddings",name:"prompt_embeds_clip"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.negative_prompt_embeds_qwen",description:"<strong>negative_prompt_embeds_qwen</strong> — Pre-computed Qwen negative prompt embeddings",name:"negative_prompt_embeds_qwen"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.negative_prompt_embeds_clip",description:"<strong>negative_prompt_embeds_clip</strong> — Pre-computed CLIP negative prompt embeddings",name:"negative_prompt_embeds_clip"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.prompt_cu_seqlens",description:"<strong>prompt_cu_seqlens</strong> — Pre-computed cumulative sequence lengths for Qwen positive prompt",name:"prompt_cu_seqlens"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.negative_prompt_cu_seqlens",description:"<strong>negative_prompt_cu_seqlens</strong> — Pre-computed cumulative sequence lengths for Qwen negative prompt",name:"negative_prompt_cu_seqlens"},{anchor:"diffusers.Kandinsky5T2VPipeline.check_inputs.callback_on_step_end_tensor_inputs",description:"<strong>callback_on_step_end_tensor_inputs</strong> — Callback tensor inputs",name:"callback_on_step_end_tensor_inputs"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky.py#L513",raiseDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<ul>
	<li><code>ValueError</code> — If inputs are invalid</li>
	</ul>
	`,raiseType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>ValueError</code></p>
	`}}),Be=new x({props:{name:"encode_prompt",anchor:"diffusers.Kandinsky5T2VPipeline.encode_prompt",parameters:[{name:"prompt",val:": typing.Union[str, typing.List[str]]"},{name:"num_videos_per_prompt",val:": int = 1"},{name:"max_sequence_length",val:": int = 512"},{name:"device",val:": typing.Optional[torch.device] = None"},{name:"dtype",val:": typing.Optional[torch.dtype] = None"}],parametersDescription:[{anchor:"diffusers.Kandinsky5T2VPipeline.encode_prompt.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>List[str]</code>) —
	Prompt to be encoded.`,name:"prompt"},{anchor:"diffusers.Kandinsky5T2VPipeline.encode_prompt.num_videos_per_prompt",description:`<strong>num_videos_per_prompt</strong> (<code>int</code>, <em>optional</em>, defaults to 1) —
	Number of videos to generate per prompt.`,name:"num_videos_per_prompt"},{anchor:"diffusers.Kandinsky5T2VPipeline.encode_prompt.max_sequence_length",description:`<strong>max_sequence_length</strong> (<code>int</code>, <em>optional</em>, defaults to 512) —
	Maximum sequence length for text encoding.`,name:"max_sequence_length"},{anchor:"diffusers.Kandinsky5T2VPipeline.encode_prompt.device",description:`<strong>device</strong> (<code>torch.device</code>, <em>optional</em>) —
	Torch device.`,name:"device"},{anchor:"diffusers.Kandinsky5T2VPipeline.encode_prompt.dtype",description:`<strong>dtype</strong> (<code>torch.dtype</code>, <em>optional</em>) —
	Torch dtype.`,name:"dtype"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky.py#L420",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<ul>
	<li>Qwen text embeddings of shape (batch_size * num_videos_per_prompt, sequence_length, embedding_dim)</li>
	<li>CLIP pooled embeddings of shape (batch_size * num_videos_per_prompt, clip_embedding_dim)</li>
	<li>Cumulative sequence lengths (<code>cu_seqlens</code>) for Qwen embeddings of shape (batch_size *
	num_videos_per_prompt + 1,)</li>
	</ul>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Tuple[torch.Tensor, torch.Tensor, torch.Tensor]</p>
	`}}),Ge=new x({props:{name:"fast_sta_nabla",anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla",parameters:[{name:"T",val:": int"},{name:"H",val:": int"},{name:"W",val:": int"},{name:"wT",val:": int = 3"},{name:"wH",val:": int = 3"},{name:"wW",val:": int = 3"},{name:"device",val:" = 'cuda'"}],parametersDescription:[{anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla.T",description:"<strong>T</strong> (int) — Number of temporal frames",name:"T"},{anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla.H",description:"<strong>H</strong> (int) — Height in latent space",name:"H"},{anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla.W",description:"<strong>W</strong> (int) — Width in latent space",name:"W"},{anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla.wT",description:"<strong>wT</strong> (int) — Temporal attention window size",name:"wT"},{anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla.wH",description:"<strong>wH</strong> (int) — Height attention window size",name:"wH"},{anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla.wW",description:"<strong>wW</strong> (int) — Width attention window size",name:"wW"},{anchor:"diffusers.Kandinsky5T2VPipeline.fast_sta_nabla.device",description:"<strong>device</strong> (str) — Device to create tensor on",name:"device"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky.py#L229",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Sparse attention mask of shape (T<em>H</em>W, T<em>H</em>W)</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>torch.Tensor</p>
	`}}),Ze=new x({props:{name:"get_sparse_params",anchor:"diffusers.Kandinsky5T2VPipeline.get_sparse_params",parameters:[{name:"sample",val:""},{name:"device",val:""}],parametersDescription:[{anchor:"diffusers.Kandinsky5T2VPipeline.get_sparse_params.sample",description:"<strong>sample</strong> (torch.Tensor) — Input sample tensor",name:"sample"},{anchor:"diffusers.Kandinsky5T2VPipeline.get_sparse_params.device",description:"<strong>device</strong> (torch.device) — Device to place tensors on",name:"device"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky.py#L264",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Dictionary containing sparse attention parameters</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Dict</p>
	`}}),We=new x({props:{name:"prepare_latents",anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents",parameters:[{name:"batch_size",val:": int"},{name:"num_channels_latents",val:": int = 16"},{name:"height",val:": int = 480"},{name:"width",val:": int = 832"},{name:"num_frames",val:": int = 81"},{name:"dtype",val:": typing.Optional[torch.dtype] = None"},{name:"device",val:": typing.Optional[torch.device] = None"},{name:"generator",val:": typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None"},{name:"latents",val:": typing.Optional[torch.Tensor] = None"}],parametersDescription:[{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.batch_size",description:"<strong>batch_size</strong> (int) — Number of videos to generate",name:"batch_size"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.num_channels_latents",description:"<strong>num_channels_latents</strong> (int) — Number of channels in latent space",name:"num_channels_latents"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.height",description:"<strong>height</strong> (int) — Height of generated video",name:"height"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.width",description:"<strong>width</strong> (int) — Width of generated video",name:"width"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.num_frames",description:"<strong>num_frames</strong> (int) — Number of frames in video",name:"num_frames"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.dtype",description:"<strong>dtype</strong> (torch.dtype) — Data type for latents",name:"dtype"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.device",description:"<strong>device</strong> (torch.device) — Device to create latents on",name:"device"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.generator",description:"<strong>generator</strong> (torch.Generator) — Random number generator",name:"generator"},{anchor:"diffusers.Kandinsky5T2VPipeline.prepare_latents.latents",description:"<strong>latents</strong> (torch.Tensor) — Pre-existing latents to use",name:"latents"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky.py#L599",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Prepared latent tensor</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>torch.Tensor</p>
	`}}),$e=new C({props:{title:"Kandinsky5I2VPipeline",local:"diffusers.Kandinsky5I2VPipeline",headingTag:"h2"}}),Pe=new x({props:{name:"class diffusers.Kandinsky5I2VPipeline",anchor:"diffusers.Kandinsky5I2VPipeline",parameters:[{name:"transformer",val:": Kandinsky5Transformer3DModel"},{name:"vae",val:": AutoencoderKLHunyuanVideo"},{name:"text_encoder",val:": Qwen2_5_VLForConditionalGeneration"},{name:"tokenizer",val:": Qwen2VLProcessor"},{name:"text_encoder_2",val:": CLIPTextModel"},{name:"tokenizer_2",val:": CLIPTokenizer"},{name:"scheduler",val:": FlowMatchEulerDiscreteScheduler"}],parametersDescription:[{anchor:"diffusers.Kandinsky5I2VPipeline.transformer",description:`<strong>transformer</strong> (<code>Kandinsky5Transformer3DModel</code>) —
	Conditional Transformer to denoise the encoded video latents.`,name:"transformer"},{anchor:"diffusers.Kandinsky5I2VPipeline.vae",description:`<strong>vae</strong> (<a href="/docs/diffusers/pr_11636/en/api/models/autoencoder_kl_hunyuan_video#diffusers.AutoencoderKLHunyuanVideo">AutoencoderKLHunyuanVideo</a>) —
	Variational Auto-Encoder Model <a href="https://huggingface.co/hunyuanvideo-community/HunyuanVideo" rel="nofollow">hunyuanvideo-community/HunyuanVideo
	(vae)</a> to encode and decode videos to and from
	latent representations.`,name:"vae"},{anchor:"diffusers.Kandinsky5I2VPipeline.text_encoder",description:`<strong>text_encoder</strong> (<code>Qwen2_5_VLForConditionalGeneration</code>) —
	Frozen text-encoder <a href="https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct" rel="nofollow">Qwen2.5-VL</a>.`,name:"text_encoder"},{anchor:"diffusers.Kandinsky5I2VPipeline.tokenizer",description:`<strong>tokenizer</strong> (<code>AutoProcessor</code>) —
	Tokenizer for Qwen2.5-VL.`,name:"tokenizer"},{anchor:"diffusers.Kandinsky5I2VPipeline.text_encoder_2",description:`<strong>text_encoder_2</strong> (<code>CLIPTextModel</code>) —
	Frozen <a href="https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel" rel="nofollow">CLIP</a>,
	specifically the <a href="https://huggingface.co/openai/clip-vit-large-patch14" rel="nofollow">clip-vit-large-patch14</a> variant.`,name:"text_encoder_2"},{anchor:"diffusers.Kandinsky5I2VPipeline.tokenizer_2",description:`<strong>tokenizer_2</strong> (<code>CLIPTokenizer</code>) —
	Tokenizer for CLIP.`,name:"tokenizer_2"},{anchor:"diffusers.Kandinsky5I2VPipeline.scheduler",description:`<strong>scheduler</strong> (<a href="/docs/diffusers/pr_11636/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 video latents.`,name:"scheduler"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky_i2v.py#L128"}}),Qe=new x({props:{name:"__call__",anchor:"diffusers.Kandinsky5I2VPipeline.__call__",parameters:[{name:"image",val:": typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor]]"},{name:"prompt",val:": typing.Union[str, typing.List[str]] = None"},{name:"negative_prompt",val:": typing.Union[str, typing.List[str], NoneType] = None"},{name:"height",val:": int = 512"},{name:"width",val:": int = 768"},{name:"num_frames",val:": int = 121"},{name:"num_inference_steps",val:": int = 50"},{name:"guidance_scale",val:": float = 5.0"},{name:"num_videos_per_prompt",val:": typing.Optional[int] = 1"},{name:"generator",val:": typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None"},{name:"latents",val:": typing.Optional[torch.Tensor] = None"},{name:"prompt_embeds_qwen",val:": typing.Optional[torch.Tensor] = None"},{name:"prompt_embeds_clip",val:": typing.Optional[torch.Tensor] = None"},{name:"negative_prompt_embeds_qwen",val:": typing.Optional[torch.Tensor] = None"},{name:"negative_prompt_embeds_clip",val:": typing.Optional[torch.Tensor] = None"},{name:"prompt_cu_seqlens",val:": typing.Optional[torch.Tensor] = None"},{name:"negative_prompt_cu_seqlens",val:": typing.Optional[torch.Tensor] = None"},{name:"output_type",val:": typing.Optional[str] = 'pil'"},{name:"return_dict",val:": bool = True"},{name:"callback_on_step_end",val:": typing.Union[typing.Callable[[int, int, typing.Dict], NoneType], diffusers.callbacks.PipelineCallback, diffusers.callbacks.MultiPipelineCallbacks, NoneType] = None"},{name:"callback_on_step_end_tensor_inputs",val:": typing.List[str] = ['latents']"},{name:"max_sequence_length",val:": int = 512"}],parametersDescription:[{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.image",description:`<strong>image</strong> (<code>PipelineImageInput</code>) —
	The input image to condition the generation on. Must be an image, a list of images or a <code>torch.Tensor</code>.`,name:"image"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>List[str]</code>, <em>optional</em>) —
	The prompt or prompts to guide the video generation. If not defined, pass <code>prompt_embeds</code> instead.`,name:"prompt"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.negative_prompt",description:`<strong>negative_prompt</strong> (<code>str</code> or <code>List[str]</code>, <em>optional</em>) —
	The prompt or prompts to avoid during video generation. If not defined, pass <code>negative_prompt_embeds</code>
	instead. Ignored when not using guidance (<code>guidance_scale</code> < <code>1</code>).`,name:"negative_prompt"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.height",description:`<strong>height</strong> (<code>int</code>, defaults to <code>512</code>) —
	The height in pixels of the generated video.`,name:"height"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.width",description:`<strong>width</strong> (<code>int</code>, defaults to <code>768</code>) —
	The width in pixels of the generated video.`,name:"width"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.num_frames",description:`<strong>num_frames</strong> (<code>int</code>, defaults to <code>121</code>) —
	The number of frames in the generated video.`,name:"num_frames"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.num_inference_steps",description:`<strong>num_inference_steps</strong> (<code>int</code>, defaults to <code>50</code>) —
	The number of denoising steps.`,name:"num_inference_steps"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.guidance_scale",description:`<strong>guidance_scale</strong> (<code>float</code>, defaults to <code>5.0</code>) —
	Guidance scale as defined in classifier-free guidance.`,name:"guidance_scale"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.num_videos_per_prompt",description:`<strong>num_videos_per_prompt</strong> (<code>int</code>, <em>optional</em>, defaults to 1) —
	The number of videos to generate per prompt.`,name:"num_videos_per_prompt"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.generator",description:`<strong>generator</strong> (<code>torch.Generator</code> or <code>List[torch.Generator]</code>, <em>optional</em>) —
	A torch generator to make generation deterministic.`,name:"generator"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.latents",description:`<strong>latents</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated noisy latents.`,name:"latents"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.prompt_embeds_qwen",description:`<strong>prompt_embeds_qwen</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated Qwen text embeddings.`,name:"prompt_embeds_qwen"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.prompt_embeds_clip",description:`<strong>prompt_embeds_clip</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated CLIP text embeddings.`,name:"prompt_embeds_clip"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.negative_prompt_embeds_qwen",description:`<strong>negative_prompt_embeds_qwen</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated Qwen negative text embeddings.`,name:"negative_prompt_embeds_qwen"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.negative_prompt_embeds_clip",description:`<strong>negative_prompt_embeds_clip</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated CLIP negative text embeddings.`,name:"negative_prompt_embeds_clip"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.prompt_cu_seqlens",description:`<strong>prompt_cu_seqlens</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated cumulative sequence lengths for Qwen positive prompt.`,name:"prompt_cu_seqlens"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.negative_prompt_cu_seqlens",description:`<strong>negative_prompt_cu_seqlens</strong> (<code>torch.Tensor</code>, <em>optional</em>) —
	Pre-generated cumulative sequence lengths for Qwen negative prompt.`,name:"negative_prompt_cu_seqlens"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.output_type",description:`<strong>output_type</strong> (<code>str</code>, <em>optional</em>, defaults to <code>"pil"</code>) —
	The output format of the generated video.`,name:"output_type"},{anchor:"diffusers.Kandinsky5I2VPipeline.__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>KandinskyPipelineOutput</code>.`,name:"return_dict"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.callback_on_step_end",description:`<strong>callback_on_step_end</strong> (<code>Callable</code>, <code>PipelineCallback</code>, <code>MultiPipelineCallbacks</code>, <em>optional</em>) —
	A function that is called at the end of each denoising step.`,name:"callback_on_step_end"},{anchor:"diffusers.Kandinsky5I2VPipeline.__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.`,name:"callback_on_step_end_tensor_inputs"},{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.max_sequence_length",description:`<strong>max_sequence_length</strong> (<code>int</code>, defaults to <code>512</code>) —
	The maximum sequence length for text encoding.`,name:"max_sequence_length"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky_i2v.py#L748",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>If <code>return_dict</code> is <code>True</code>, <code>KandinskyPipelineOutput</code> is returned, otherwise a <code>tuple</code> is returned
	where the first element is a list with the generated videos.</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>~KandinskyPipelineOutput</code> or <code>tuple</code></p>
	`}}),A=new Ys({props:{anchor:"diffusers.Kandinsky5I2VPipeline.__call__.example",$$slots:{default:[ta]},$$scope:{ctx:Se}}}),Ke=new x({props:{name:"check_inputs",anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs",parameters:[{name:"prompt",val:""},{name:"negative_prompt",val:""},{name:"image",val:""},{name:"height",val:""},{name:"width",val:""},{name:"prompt_embeds_qwen",val:" = None"},{name:"prompt_embeds_clip",val:" = None"},{name:"negative_prompt_embeds_qwen",val:" = None"},{name:"negative_prompt_embeds_clip",val:" = None"},{name:"prompt_cu_seqlens",val:" = None"},{name:"negative_prompt_cu_seqlens",val:" = None"},{name:"callback_on_step_end_tensor_inputs",val:" = None"},{name:"max_sequence_length",val:" = None"}],parametersDescription:[{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.prompt",description:"<strong>prompt</strong> — Input prompt",name:"prompt"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.negative_prompt",description:"<strong>negative_prompt</strong> — Negative prompt for guidance",name:"negative_prompt"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.image",description:"<strong>image</strong> — Input image for conditioning",name:"image"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.height",description:"<strong>height</strong> — Video height",name:"height"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.width",description:"<strong>width</strong> — Video width",name:"width"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.prompt_embeds_qwen",description:"<strong>prompt_embeds_qwen</strong> — Pre-computed Qwen prompt embeddings",name:"prompt_embeds_qwen"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.prompt_embeds_clip",description:"<strong>prompt_embeds_clip</strong> — Pre-computed CLIP prompt embeddings",name:"prompt_embeds_clip"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.negative_prompt_embeds_qwen",description:"<strong>negative_prompt_embeds_qwen</strong> — Pre-computed Qwen negative prompt embeddings",name:"negative_prompt_embeds_qwen"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.negative_prompt_embeds_clip",description:"<strong>negative_prompt_embeds_clip</strong> — Pre-computed CLIP negative prompt embeddings",name:"negative_prompt_embeds_clip"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.prompt_cu_seqlens",description:"<strong>prompt_cu_seqlens</strong> — Pre-computed cumulative sequence lengths for Qwen positive prompt",name:"prompt_cu_seqlens"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.negative_prompt_cu_seqlens",description:"<strong>negative_prompt_cu_seqlens</strong> — Pre-computed cumulative sequence lengths for Qwen negative prompt",name:"negative_prompt_cu_seqlens"},{anchor:"diffusers.Kandinsky5I2VPipeline.check_inputs.callback_on_step_end_tensor_inputs",description:"<strong>callback_on_step_end_tensor_inputs</strong> — Callback tensor inputs",name:"callback_on_step_end_tensor_inputs"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky_i2v.py#L545",raiseDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<ul>
	<li><code>ValueError</code> — If inputs are invalid</li>
	</ul>
	`,raiseType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p><code>ValueError</code></p>
	`}}),Ee=new x({props:{name:"encode_prompt",anchor:"diffusers.Kandinsky5I2VPipeline.encode_prompt",parameters:[{name:"prompt",val:": typing.Union[str, typing.List[str]]"},{name:"num_videos_per_prompt",val:": int = 1"},{name:"max_sequence_length",val:": int = 512"},{name:"device",val:": typing.Optional[torch.device] = None"},{name:"dtype",val:": typing.Optional[torch.dtype] = None"}],parametersDescription:[{anchor:"diffusers.Kandinsky5I2VPipeline.encode_prompt.prompt",description:`<strong>prompt</strong> (<code>str</code> or <code>List[str]</code>) —
	Prompt to be encoded.`,name:"prompt"},{anchor:"diffusers.Kandinsky5I2VPipeline.encode_prompt.num_videos_per_prompt",description:`<strong>num_videos_per_prompt</strong> (<code>int</code>, <em>optional</em>, defaults to 1) —
	Number of videos to generate per prompt.`,name:"num_videos_per_prompt"},{anchor:"diffusers.Kandinsky5I2VPipeline.encode_prompt.max_sequence_length",description:`<strong>max_sequence_length</strong> (<code>int</code>, <em>optional</em>, defaults to 512) —
	Maximum sequence length for text encoding.`,name:"max_sequence_length"},{anchor:"diffusers.Kandinsky5I2VPipeline.encode_prompt.device",description:`<strong>device</strong> (<code>torch.device</code>, <em>optional</em>) —
	Torch device.`,name:"device"},{anchor:"diffusers.Kandinsky5I2VPipeline.encode_prompt.dtype",description:`<strong>dtype</strong> (<code>torch.dtype</code>, <em>optional</em>) —
	Torch dtype.`,name:"dtype"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky_i2v.py#L454",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<ul>
	<li>Qwen text embeddings of shape (batch_size * num_videos_per_prompt, sequence_length, embedding_dim)</li>
	<li>CLIP pooled embeddings of shape (batch_size * num_videos_per_prompt, clip_embedding_dim)</li>
	<li>Cumulative sequence lengths (<code>cu_seqlens</code>) for Qwen embeddings of shape (batch_size *
	num_videos_per_prompt + 1,)</li>
	</ul>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Tuple[torch.Tensor, torch.Tensor, torch.Tensor]</p>
	`}}),Ne=new x({props:{name:"fast_sta_nabla",anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla",parameters:[{name:"T",val:": int"},{name:"H",val:": int"},{name:"W",val:": int"},{name:"wT",val:": int = 3"},{name:"wH",val:": int = 3"},{name:"wW",val:": int = 3"},{name:"device",val:" = 'cuda'"}],parametersDescription:[{anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla.T",description:"<strong>T</strong> (int) — Number of temporal frames",name:"T"},{anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla.H",description:"<strong>H</strong> (int) — Height in latent space",name:"H"},{anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla.W",description:"<strong>W</strong> (int) — Width in latent space",name:"W"},{anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla.wT",description:"<strong>wT</strong> (int) — Temporal attention window size",name:"wT"},{anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla.wH",description:"<strong>wH</strong> (int) — Height attention window size",name:"wH"},{anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla.wW",description:"<strong>wW</strong> (int) — Width attention window size",name:"wW"},{anchor:"diffusers.Kandinsky5I2VPipeline.fast_sta_nabla.device",description:"<strong>device</strong> (str) — Device to create tensor on",name:"device"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky_i2v.py#L226",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Sparse attention mask of shape (T<em>H</em>W, T<em>H</em>W)</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>torch.Tensor</p>
	`}}),Xe=new x({props:{name:"get_sparse_params",anchor:"diffusers.Kandinsky5I2VPipeline.get_sparse_params",parameters:[{name:"sample",val:""},{name:"device",val:""}],parametersDescription:[{anchor:"diffusers.Kandinsky5I2VPipeline.get_sparse_params.sample",description:"<strong>sample</strong> (torch.Tensor) — Input sample tensor",name:"sample"},{anchor:"diffusers.Kandinsky5I2VPipeline.get_sparse_params.device",description:"<strong>device</strong> (torch.device) — Device to place tensors on",name:"device"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky_i2v.py#L261",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Dictionary containing sparse attention parameters</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Dict</p>
	`}}),Fe=new x({props:{name:"prepare_latents",anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents",parameters:[{name:"image",val:": typing.Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, typing.List[PIL.Image.Image], typing.List[numpy.ndarray], typing.List[torch.Tensor]]"},{name:"batch_size",val:": int"},{name:"num_channels_latents",val:": int = 16"},{name:"height",val:": int = 480"},{name:"width",val:": int = 832"},{name:"num_frames",val:": int = 81"},{name:"dtype",val:": typing.Optional[torch.dtype] = None"},{name:"device",val:": typing.Optional[torch.device] = None"},{name:"generator",val:": typing.Union[torch._C.Generator, typing.List[torch._C.Generator], NoneType] = None"},{name:"latents",val:": typing.Optional[torch.Tensor] = None"}],parametersDescription:[{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.image",description:"<strong>image</strong> (PipelineImageInput) — Input image to condition the generation on",name:"image"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.batch_size",description:"<strong>batch_size</strong> (int) — Number of videos to generate",name:"batch_size"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.num_channels_latents",description:"<strong>num_channels_latents</strong> (int) — Number of channels in latent space",name:"num_channels_latents"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.height",description:"<strong>height</strong> (int) — Height of generated video",name:"height"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.width",description:"<strong>width</strong> (int) — Width of generated video",name:"width"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.num_frames",description:"<strong>num_frames</strong> (int) — Number of frames in video",name:"num_frames"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.dtype",description:"<strong>dtype</strong> (torch.dtype) — Data type for latents",name:"dtype"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.device",description:"<strong>device</strong> (torch.device) — Device to create latents on",name:"device"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.generator",description:"<strong>generator</strong> (torch.Generator) — Random number generator",name:"generator"},{anchor:"diffusers.Kandinsky5I2VPipeline.prepare_latents.latents",description:"<strong>latents</strong> (torch.Tensor) — Pre-existing latents to use",name:"latents"}],source:"https://github.com/huggingface/diffusers/blob/vr_11636/src/diffusers/pipelines/kandinsky5/pipeline_kandinsky_i2v.py#L636",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>Prepared latent tensor with first frame as encoded image</p>
	`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


	<p>torch.Tensor</p>
	`}}),Le=new C({props:{title:"Citation",local:"citation",headingTag:"h2"}}),Re=new q({props:{code:"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",highlighted:`<span class="hljs-comment">@misc{kandinsky2025,</span>
	author = {Alexander Belykh <span class="hljs-keyword">and</span> Alexander Varlamov <span class="hljs-keyword">and</span> Alexey Letunovskiy <span class="hljs-keyword">and</span> Anastasia Aliaskina <span class="hljs-keyword">and</span> Anastasia Maltseva <span class="hljs-keyword">and</span> Anastasiia Kargapoltseva <span class="hljs-keyword">and</span> Andrey Shutkin <span class="hljs-keyword">and</span> Anna Averchenkova <span class="hljs-keyword">and</span> Anna Dmitrienko <span class="hljs-keyword">and</span> Bulat Akhmatov <span class="hljs-keyword">and</span> Denis Dimitrov <span class="hljs-keyword">and</span> Denis Koposov <span class="hljs-keyword">and</span> Denis Parkhomenko <span class="hljs-keyword">and</span> Dmitrii <span class="hljs-keyword">and</span> Ilya Vasiliev <span class="hljs-keyword">and</span> Ivan Kirillov <span class="hljs-keyword">and</span> Julia Agafonova <span class="hljs-keyword">and</span> Kirill Chernyshev <span class="hljs-keyword">and</span> Kormilitsyn Semen <span class="hljs-keyword">and</span> Lev Novitskiy <span class="hljs-keyword">and</span> Maria Kovaleva <span class="hljs-keyword">and</span> Mikhail Mamaev <span class="hljs-keyword">and</span> Mikhailov <span class="hljs-keyword">and</span> Nikita Kiselev <span class="hljs-keyword">and</span> Nikita Osterov <span class="hljs-keyword">and</span> Nikolai Gerasimenko <span class="hljs-keyword">and</span> Nikolai Vaulin <span class="hljs-keyword">and</span> Olga Kim <span class="hljs-keyword">and</span> Olga Vdovchenko <span class="hljs-keyword">and</span> Polina Gavrilova <span class="hljs-keyword">and</span> Polina Mikhailova <span class="hljs-keyword">and</span> Tatiana Nikulina <span class="hljs-keyword">and</span> Viacheslav Vasilev <span class="hljs-keyword">and</span> Vladimir Arkhipkin <span class="hljs-keyword">and</span> Vladimir Korviakov <span class="hljs-keyword">and</span> Vladimir Polovnikov <span class="hljs-keyword">and</span> Yury Kolabushin},
	title = {Kandinsky <span class="hljs-number">5</span>.<span class="hljs-number">0</span>: A family of diffusion models for Video & Image generation},
	howpublished = {\\url{https:<span class="hljs-comment">//github.com/kandinskylab/Kandinsky-5}},</span>
	year = <span class="hljs-number">2025</span>
	}`,wrap:!1}}),Ye=new ea({props:{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/pipelines/kandinsky5_video.md"}}),{c(){f=o("meta"),G=s(),J=o("p"),T=s(),d(v.$$.fragment),h=s(),d(j.$$.fragment),xn=s(),D=o("p"),D.innerHTML=as,In=s(),O=o("p"),O.textContent=is,Vn=s(),ee=o("p"),ee.textContent=ls,Bn=s(),ne=o("p"),ne.textContent=os,Gn=s(),te=o("ul"),te.innerHTML=rs,Zn=s(),se=o("p"),se.innerHTML=ps,Wn=s(),Y=o("blockquote"),Y.innerHTML=ds,$n=s(),d(ae.$$.fragment),Pn=s(),ie=o("p"),ie.textContent=cs,Qn=s(),le=o("table"),le.innerHTML=ms,Kn=s(),oe=o("p"),oe.innerHTML=us,En=s(),d(re.$$.fragment),Nn=s(),d(pe.$$.fragment),Xn=s(),d(de.$$.fragment),Fn=s(),ce=o("p"),ce.innerHTML=ys,Ln=s(),d(me.$$.fragment),Rn=s(),d(ue.$$.fragment),Yn=s(),d(ye.$$.fragment),Sn=s(),d(ge.$$.fragment),Hn=s(),he=o("p"),he.innerHTML=gs,An=s(),d(fe.$$.fragment),zn=s(),d(_e.$$.fragment),qn=s(),Me=o("p"),Me.innerHTML=hs,Dn=s(),d(be.$$.fragment),On=s(),d(Te.$$.fragment),et=s(),ke=o("p"),ke.innerHTML=fs,nt=s(),d(we.$$.fragment),tt=s(),d(ve.$$.fragment),st=s(),I=o("table"),I.innerHTML=_s,at=s(),d(Ue.$$.fragment),it=s(),Je=o("p"),Je.innerHTML=Ms,lt=s(),V=o("table"),V.innerHTML=bs,ot=s(),d(je.$$.fragment),rt=s(),B=o("table"),B.innerHTML=Ts,pt=s(),d(Ce.$$.fragment),dt=s(),_=o("div"),d(xe.$$.fragment),bt=s(),He=o("p"),He.textContent=ks,Tt=s(),Ae=o("p"),Ae.innerHTML=ws,kt=s(),Z=o("div"),d(Ie.$$.fragment),wt=s(),ze=o("p"),ze.textContent=vs,vt=s(),d(S.$$.fragment),Ut=s(),H=o("div"),d(Ve.$$.fragment),Jt=s(),qe=o("p"),qe.textContent=Us,jt=s(),W=o("div"),d(Be.$$.fragment),Ct=s(),De=o("p"),De.textContent=Js,xt=s(),Oe=o("p"),Oe.textContent=js,It=s(),$=o("div"),d(Ge.$$.fragment),Vt=s(),en=o("p"),en.textContent=Cs,Bt=s(),nn=o("p"),nn.textContent=xs,Gt=s(),P=o("div"),d(Ze.$$.fragment),Zt=s(),tn=o("p"),tn.textContent=Is,Wt=s(),sn=o("p"),sn.textContent=Vs,$t=s(),Q=o("div"),d(We.$$.fragment),Pt=s(),an=o("p"),an.textContent=Bs,Qt=s(),ln=o("p"),ln.textContent=Gs,ct=s(),d($e.$$.fragment),mt=s(),M=o("div"),d(Pe.$$.fragment),Kt=s(),on=o("p"),on.textContent=Zs,Et=s(),rn=o("p"),rn.innerHTML=Ws,Nt=s(),K=o("div"),d(Qe.$$.fragment),Xt=s(),pn=o("p"),pn.textContent=$s,Ft=s(),d(A.$$.fragment),Lt=s(),z=o("div"),d(Ke.$$.fragment),Rt=s(),dn=o("p"),dn.textContent=Ps,Yt=s(),E=o("div"),d(Ee.$$.fragment),St=s(),cn=o("p"),cn.textContent=Qs,Ht=s(),mn=o("p"),mn.textContent=Ks,At=s(),N=o("div"),d(Ne.$$.fragment),zt=s(),un=o("p"),un.textContent=Es,qt=s(),yn=o("p"),yn.textContent=Ns,Dt=s(),X=o("div"),d(Xe.$$.fragment),Ot=s(),gn=o("p"),gn.textContent=Xs,es=s(),hn=o("p"),hn.textContent=Fs,ns=s(),F=o("div"),d(Fe.$$.fragment),ts=s(),fn=o("p"),fn.textContent=Ls,ss=s(),_n=o("p"),_n.textContent=Rs,ut=s(),d(Le.$$.fragment),yt=s(),d(Re.$$.fragment),gt=s(),d(Ye.$$.fragment),ht=s(),Cn=o("p"),this.h()},l(e){const n=Ds("svelte-u9bgzb",document.head);f=r(n,"META",{name:!0,content:!0}),n.forEach(t),G=a(e),J=r(e,"P",{}),U(J).forEach(t),T=a(e),c(v.$$.fragment,e),h=a(e),c(j.$$.fragment,e),xn=a(e),D=r(e,"P",{"data-svelte-h":!0}),p(D)!=="svelte-1o4s5w8"&&(D.innerHTML=as),In=a(e),O=r(e,"P",{"data-svelte-h":!0}),p(O)!=="svelte-1i8pwt3"&&(O.textContent=is),Vn=a(e),ee=r(e,"P",{"data-svelte-h":!0}),p(ee)!=="svelte-1ta4ctk"&&(ee.textContent=ls),Bn=a(e),ne=r(e,"P",{"data-svelte-h":!0}),p(ne)!=="svelte-cwew0b"&&(ne.textContent=os),Gn=a(e),te=r(e,"UL",{"data-svelte-h":!0}),p(te)!=="svelte-1vr585w"&&(te.innerHTML=rs),Zn=a(e),se=r(e,"P",{"data-svelte-h":!0}),p(se)!=="svelte-729g83"&&(se.innerHTML=ps),Wn=a(e),Y=r(e,"BLOCKQUOTE",{class:!0,"data-svelte-h":!0}),p(Y)!=="svelte-131mg47"&&(Y.innerHTML=ds),$n=a(e),c(ae.$$.fragment,e),Pn=a(e),ie=r(e,"P",{"data-svelte-h":!0}),p(ie)!=="svelte-x278iq"&&(ie.textContent=cs),Qn=a(e),le=r(e,"TABLE",{"data-svelte-h":!0}),p(le)!=="svelte-cdx4w4"&&(le.innerHTML=ms),Kn=a(e),oe=r(e,"P",{"data-svelte-h":!0}),p(oe)!=="svelte-d3wi78"&&(oe.innerHTML=us),En=a(e),c(re.$$.fragment,e),Nn=a(e),c(pe.$$.fragment,e),Xn=a(e),c(de.$$.fragment,e),Fn=a(e),ce=r(e,"P",{"data-svelte-h":!0}),p(ce)!=="svelte-mxui5h"&&(ce.innerHTML=ys),Ln=a(e),c(me.$$.fragment,e),Rn=a(e),c(ue.$$.fragment,e),Yn=a(e),c(ye.$$.fragment,e),Sn=a(e),c(ge.$$.fragment,e),Hn=a(e),he=r(e,"P",{"data-svelte-h":!0}),p(he)!=="svelte-1kr0lxg"&&(he.innerHTML=gs),An=a(e),c(fe.$$.fragment,e),zn=a(e),c(_e.$$.fragment,e),qn=a(e),Me=r(e,"P",{"data-svelte-h":!0}),p(Me)!=="svelte-188xe6v"&&(Me.innerHTML=hs),Dn=a(e),c(be.$$.fragment,e),On=a(e),c(Te.$$.fragment,e),et=a(e),ke=r(e,"P",{"data-svelte-h":!0}),p(ke)!=="svelte-mxui5h"&&(ke.innerHTML=fs),nt=a(e),c(we.$$.fragment,e),tt=a(e),c(ve.$$.fragment,e),st=a(e),I=r(e,"TABLE",{border:!0,style:!0,"data-svelte-h":!0}),p(I)!=="svelte-16ixh8u"&&(I.innerHTML=_s),at=a(e),c(Ue.$$.fragment,e),it=a(e),Je=r(e,"P",{"data-svelte-h":!0}),p(Je)!=="svelte-9k6zim"&&(Je.innerHTML=Ms),lt=a(e),V=r(e,"TABLE",{border:!0,style:!0,"data-svelte-h":!0}),p(V)!=="svelte-1lcaor5"&&(V.innerHTML=bs),ot=a(e),c(je.$$.fragment,e),rt=a(e),B=r(e,"TABLE",{border:!0,style:!0,"data-svelte-h":!0}),p(B)!=="svelte-s1hzum"&&(B.innerHTML=Ts),pt=a(e),c(Ce.$$.fragment,e),dt=a(e),_=r(e,"DIV",{class:!0});var b=U(_);c(xe.$$.fragment,b),bt=a(b),He=r(b,"P",{"data-svelte-h":!0}),p(He)!=="svelte-1hnedav"&&(He.textContent=ks),Tt=a(b),Ae=r(b,"P",{"data-svelte-h":!0}),p(Ae)!=="svelte-1w7teon"&&(Ae.innerHTML=ws),kt=a(b),Z=r(b,"DIV",{class:!0});var L=U(Z);c(Ie.$$.fragment,L),wt=a(L),ze=r(L,"P",{"data-svelte-h":!0}),p(ze)!=="svelte-50j04k"&&(ze.textContent=vs),vt=a(L),c(S.$$.fragment,L),L.forEach(t),Ut=a(b),H=r(b,"DIV",{class:!0});var _t=U(H);c(Ve.$$.fragment,_t),Jt=a(_t),qe=r(_t,"P",{"data-svelte-h":!0}),p(qe)!=="svelte-11b6rmu"&&(qe.textContent=Us),_t.forEach(t),jt=a(b),W=r(b,"DIV",{class:!0});var Mn=U(W);c(Be.$$.fragment,Mn),Ct=a(Mn),De=r(Mn,"P",{"data-svelte-h":!0}),p(De)!=="svelte-1vxta4j"&&(De.textContent=Js),xt=a(Mn),Oe=r(Mn,"P",{"data-svelte-h":!0}),p(Oe)!=="svelte-1jhtoxo"&&(Oe.textContent=js),Mn.forEach(t),It=a(b),$=r(b,"DIV",{class:!0});var bn=U($);c(Ge.$$.fragment,bn),Vt=a(bn),en=r(bn,"P",{"data-svelte-h":!0}),p(en)!=="svelte-prp4gp"&&(en.textContent=Cs),Bt=a(bn),nn=r(bn,"P",{"data-svelte-h":!0}),p(nn)!=="svelte-1i366n7"&&(nn.textContent=xs),bn.forEach(t),Gt=a(b),P=r(b,"DIV",{class:!0});var Tn=U(P);c(Ze.$$.fragment,Tn),Zt=a(Tn),tn=r(Tn,"P",{"data-svelte-h":!0}),p(tn)!=="svelte-11b03ij"&&(tn.textContent=Is),Wt=a(Tn),sn=r(Tn,"P",{"data-svelte-h":!0}),p(sn)!=="svelte-1q4t3h1"&&(sn.textContent=Vs),Tn.forEach(t),$t=a(b),Q=r(b,"DIV",{class:!0});var kn=U(Q);c(We.$$.fragment,kn),Pt=a(kn),an=r(kn,"P",{"data-svelte-h":!0}),p(an)!=="svelte-1auko1i"&&(an.textContent=Bs),Qt=a(kn),ln=r(kn,"P",{"data-svelte-h":!0}),p(ln)!=="svelte-1bvour2"&&(ln.textContent=Gs),kn.forEach(t),b.forEach(t),ct=a(e),c($e.$$.fragment,e),mt=a(e),M=r(e,"DIV",{class:!0});var w=U(M);c(Pe.$$.fragment,w),Kt=a(w),on=r(w,"P",{"data-svelte-h":!0}),p(on)!=="svelte-19chvdh"&&(on.textContent=Zs),Et=a(w),rn=r(w,"P",{"data-svelte-h":!0}),p(rn)!=="svelte-1w7teon"&&(rn.innerHTML=Ws),Nt=a(w),K=r(w,"DIV",{class:!0});var wn=U(K);c(Qe.$$.fragment,wn),Xt=a(wn),pn=r(wn,"P",{"data-svelte-h":!0}),p(pn)!=="svelte-en0orh"&&(pn.textContent=$s),Ft=a(wn),c(A.$$.fragment,wn),wn.forEach(t),Lt=a(w),z=r(w,"DIV",{class:!0});var Mt=U(z);c(Ke.$$.fragment,Mt),Rt=a(Mt),dn=r(Mt,"P",{"data-svelte-h":!0}),p(dn)!=="svelte-11b6rmu"&&(dn.textContent=Ps),Mt.forEach(t),Yt=a(w),E=r(w,"DIV",{class:!0});var vn=U(E);c(Ee.$$.fragment,vn),St=a(vn),cn=r(vn,"P",{"data-svelte-h":!0}),p(cn)!=="svelte-1vxta4j"&&(cn.textContent=Qs),Ht=a(vn),mn=r(vn,"P",{"data-svelte-h":!0}),p(mn)!=="svelte-1jhtoxo"&&(mn.textContent=Ks),vn.forEach(t),At=a(w),N=r(w,"DIV",{class:!0});var Un=U(N);c(Ne.$$.fragment,Un),zt=a(Un),un=r(Un,"P",{"data-svelte-h":!0}),p(un)!=="svelte-prp4gp"&&(un.textContent=Es),qt=a(Un),yn=r(Un,"P",{"data-svelte-h":!0}),p(yn)!=="svelte-1i366n7"&&(yn.textContent=Ns),Un.forEach(t),Dt=a(w),X=r(w,"DIV",{class:!0});var Jn=U(X);c(Xe.$$.fragment,Jn),Ot=a(Jn),gn=r(Jn,"P",{"data-svelte-h":!0}),p(gn)!=="svelte-11b03ij"&&(gn.textContent=Xs),es=a(Jn),hn=r(Jn,"P",{"data-svelte-h":!0}),p(hn)!=="svelte-1q4t3h1"&&(hn.textContent=Fs),Jn.forEach(t),ns=a(w),F=r(w,"DIV",{class:!0});var jn=U(F);c(Fe.$$.fragment,jn),ts=a(jn),fn=r(jn,"P",{"data-svelte-h":!0}),p(fn)!=="svelte-1qln60k"&&(fn.textContent=Ls),ss=a(jn),_n=r(jn,"P",{"data-svelte-h":!0}),p(_n)!=="svelte-q4pd1u"&&(_n.textContent=Rs),jn.forEach(t),w.forEach(t),ut=a(e),c(Le.$$.fragment,e),yt=a(e),c(Re.$$.fragment,e),gt=a(e),c(Ye.$$.fragment,e),ht=a(e),Cn=r(e,"P",{}),U(Cn).forEach(t),this.h()},h(){k(f,"name","hf:doc:metadata"),k(f,"content",aa),k(Y,"class","tip"),k(I,"border","0"),R(I,"width","200"),R(I,"text-align","left"),R(I,"margin-top","20px"),k(V,"border","0"),R(V,"width","400"),R(V,"text-align","left"),R(V,"margin-top","20px"),k(B,"border","0"),R(B,"width","400"),R(B,"text-align","left"),R(B,"margin-top","20px"),k(Z,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(H,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(W,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k($,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(P,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(Q,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(_,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(K,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(z,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(E,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(N,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(X,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(F,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),k(M,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8")},m(e,n){i(document.head,f),l(e,G,n),l(e,J,n),l(e,T,n),m(v,e,n),l(e,h,n),m(j,e,n),l(e,xn,n),l(e,D,n),l(e,In,n),l(e,O,n),l(e,Vn,n),l(e,ee,n),l(e,Bn,n),l(e,ne,n),l(e,Gn,n),l(e,te,n),l(e,Zn,n),l(e,se,n),l(e,Wn,n),l(e,Y,n),l(e,$n,n),m(ae,e,n),l(e,Pn,n),l(e,ie,n),l(e,Qn,n),l(e,le,n),l(e,Kn,n),l(e,oe,n),l(e,En,n),m(re,e,n),l(e,Nn,n),m(pe,e,n),l(e,Xn,n),m(de,e,n),l(e,Fn,n),l(e,ce,n),l(e,Ln,n),m(me,e,n),l(e,Rn,n),m(ue,e,n),l(e,Yn,n),m(ye,e,n),l(e,Sn,n),m(ge,e,n),l(e,Hn,n),l(e,he,n),l(e,An,n),m(fe,e,n),l(e,zn,n),m(_e,e,n),l(e,qn,n),l(e,Me,n),l(e,Dn,n),m(be,e,n),l(e,On,n),m(Te,e,n),l(e,et,n),l(e,ke,n),l(e,nt,n),m(we,e,n),l(e,tt,n),m(ve,e,n),l(e,st,n),l(e,I,n),l(e,at,n),m(Ue,e,n),l(e,it,n),l(e,Je,n),l(e,lt,n),l(e,V,n),l(e,ot,n),m(je,e,n),l(e,rt,n),l(e,B,n),l(e,pt,n),m(Ce,e,n),l(e,dt,n),l(e,_,n),m(xe,_,null),i(_,bt),i(_,He),i(_,Tt),i(_,Ae),i(_,kt),i(_,Z),m(Ie,Z,null),i(Z,wt),i(Z,ze),i(Z,vt),m(S,Z,null),i(_,Ut),i(_,H),m(Ve,H,null),i(H,Jt),i(H,qe),i(_,jt),i(_,W),m(Be,W,null),i(W,Ct),i(W,De),i(W,xt),i(W,Oe),i(_,It),i(_,$),m(Ge,$,null),i($,Vt),i($,en),i($,Bt),i($,nn),i(_,Gt),i(_,P),m(Ze,P,null),i(P,Zt),i(P,tn),i(P,Wt),i(P,sn),i(_,$t),i(_,Q),m(We,Q,null),i(Q,Pt),i(Q,an),i(Q,Qt),i(Q,ln),l(e,ct,n),m($e,e,n),l(e,mt,n),l(e,M,n),m(Pe,M,null),i(M,Kt),i(M,on),i(M,Et),i(M,rn),i(M,Nt),i(M,K),m(Qe,K,null),i(K,Xt),i(K,pn),i(K,Ft),m(A,K,null),i(M,Lt),i(M,z),m(Ke,z,null),i(z,Rt),i(z,dn),i(M,Yt),i(M,E),m(Ee,E,null),i(E,St),i(E,cn),i(E,Ht),i(E,mn),i(M,At),i(M,N),m(Ne,N,null),i(N,zt),i(N,un),i(N,qt),i(N,yn),i(M,Dt),i(M,X),m(Xe,X,null),i(X,Ot),i(X,gn),i(X,es),i(X,hn),i(M,ns),i(M,F),m(Fe,F,null),i(F,ts),i(F,fn),i(F,ss),i(F,_n),l(e,ut,n),m(Le,e,n),l(e,yt,n),m(Re,e,n),l(e,gt,n),m(Ye,e,n),l(e,ht,n),l(e,Cn,n),ft=!0},p(e,[n]){const b={};n&2&&(b.$$scope={dirty:n,ctx:e}),S.$set(b);const L={};n&2&&(L.$$scope={dirty:n,ctx:e}),A.$set(L)},i(e){ft\|\|(u(v.$$.fragment,e),u(j.$$.fragment,e),u(ae.$$.fragment,e),u(re.$$.fragment,e),u(pe.$$.fragment,e),u(de.$$.fragment,e),u(me.$$.fragment,e),u(ue.$$.fragment,e),u(ye.$$.fragment,e),u(ge.$$.fragment,e),u(fe.$$.fragment,e),u(_e.$$.fragment,e),u(be.$$.fragment,e),u(Te.$$.fragment,e),u(we.$$.fragment,e),u(ve.$$.fragment,e),u(Ue.$$.fragment,e),u(je.$$.fragment,e),u(Ce.$$.fragment,e),u(xe.$$.fragment,e),u(Ie.$$.fragment,e),u(S.$$.fragment,e),u(Ve.$$.fragment,e),u(Be.$$.fragment,e),u(Ge.$$.fragment,e),u(Ze.$$.fragment,e),u(We.$$.fragment,e),u($e.$$.fragment,e),u(Pe.$$.fragment,e),u(Qe.$$.fragment,e),u(A.$$.fragment,e),u(Ke.$$.fragment,e),u(Ee.$$.fragment,e),u(Ne.$$.fragment,e),u(Xe.$$.fragment,e),u(Fe.$$.fragment,e),u(Le.$$.fragment,e),u(Re.$$.fragment,e),u(Ye.$$.fragment,e),ft=!0)},o(e){y(v.$$.fragment,e),y(j.$$.fragment,e),y(ae.$$.fragment,e),y(re.$$.fragment,e),y(pe.$$.fragment,e),y(de.$$.fragment,e),y(me.$$.fragment,e),y(ue.$$.fragment,e),y(ye.$$.fragment,e),y(ge.$$.fragment,e),y(fe.$$.fragment,e),y(_e.$$.fragment,e),y(be.$$.fragment,e),y(Te.$$.fragment,e),y(we.$$.fragment,e),y(ve.$$.fragment,e),y(Ue.$$.fragment,e),y(je.$$.fragment,e),y(Ce.$$.fragment,e),y(xe.$$.fragment,e),y(Ie.$$.fragment,e),y(S.$$.fragment,e),y(Ve.$$.fragment,e),y(Be.$$.fragment,e),y(Ge.$$.fragment,e),y(Ze.$$.fragment,e),y(We.$$.fragment,e),y($e.$$.fragment,e),y(Pe.$$.fragment,e),y(Qe.$$.fragment,e),y(A.$$.fragment,e),y(Ke.$$.fragment,e),y(Ee.$$.fragment,e),y(Ne.$$.fragment,e),y(Xe.$$.fragment,e),y(Fe.$$.fragment,e),y(Le.$$.fragment,e),y(Re.$$.fragment,e),y(Ye.$$.fragment,e),ft=!1},d(e){e&&(t(G),t(J),t(T),t(h),t(xn),t(D),t(In),t(O),t(Vn),t(ee),t(Bn),t(ne),t(Gn),t(te),t(Zn),t(se),t(Wn),t(Y),t($n),t(Pn),t(ie),t(Qn),t(le),t(Kn),t(oe),t(En),t(Nn),t(Xn),t(Fn),t(ce),t(Ln),t(Rn),t(Yn),t(Sn),t(Hn),t(he),t(An),t(zn),t(qn),t(Me),t(Dn),t(On),t(et),t(ke),t(nt),t(tt),t(st),t(I),t(at),t(it),t(Je),t(lt),t(V),t(ot),t(rt),t(B),t(pt),t(dt),t(_),t(ct),t(mt),t(M),t(ut),t(yt),t(gt),t(ht),t(Cn)),t(f),g(v,e),g(j,e),g(ae,e),g(re,e),g(pe,e),g(de,e),g(me,e),g(ue,e),g(ye,e),g(ge,e),g(fe,e),g(_e,e),g(be,e),g(Te,e),g(we,e),g(ve,e),g(Ue,e),g(je,e),g(Ce,e),g(xe),g(Ie),g(S),g(Ve),g(Be),g(Ge),g(Ze),g(We),g($e,e),g(Pe),g(Qe),g(A),g(Ke),g(Ee),g(Ne),g(Xe),g(Fe),g(Le,e),g(Re,e),g(Ye,e)}}}const aa='{"title":"Kandinsky 5.0 Video","local":"kandinsky-50-video","sections":[{"title":"Available Models","local":"available-models","sections":[],"depth":2},{"title":"Usage Examples","local":"usage-examples","sections":[{"title":"Basic Text-to-Video Generation","local":"basic-text-to-video-generation","sections":[{"title":"Pro","local":"pro","sections":[],"depth":4},{"title":"Lite","local":"lite","sections":[],"depth":4}],"depth":3},{"title":"10 second Models","local":"10-second-models","sections":[],"depth":3},{"title":"Diffusion Distilled model","local":"diffusion-distilled-model","sections":[],"depth":3},{"title":"Basic Image-to-Video Generation","local":"basic-image-to-video-generation","sections":[],"depth":3}],"depth":2},{"title":"Kandinsky 5.0 Pro Side-by-Side evaluation","local":"kandinsky-50-pro-side-by-side-evaluation","sections":[],"depth":2},{"title":"Kandinsky 5.0 Lite Side-by-Side evaluation","local":"kandinsky-50-lite-side-by-side-evaluation","sections":[],"depth":2},{"title":"Kandinsky 5.0 Lite Distill Side-by-Side evaluation","local":"kandinsky-50-lite-distill-side-by-side-evaluation","sections":[],"depth":2},{"title":"Kandinsky5T2VPipeline","local":"diffusers.Kandinsky5T2VPipeline","sections":[],"depth":2},{"title":"Kandinsky5I2VPipeline","local":"diffusers.Kandinsky5I2VPipeline","sections":[],"depth":2},{"title":"Citation","local":"citation","sections":[],"depth":2}],"depth":1}';function ia(Se){return As(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class ua extends zs{constructor(f){super(),qs(this,f,ia,sa,Hs,{})}}export{ua as component};

Xet Storage Details

Size:: 102 kB
Xet hash:: 6949719a31139c4d6ebcacb035bf904999c1807090f767e09d27f5a19c83022e

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