<h3 id="introduction">🎬 Introduction</h3>Welcome to a powerhouse ComfyUI workflow designed to unlock the incredible potential of the Wan 2.2 14B Sound-to-Video (S2V) model. This isn't just a simple implementation; it's a comprehensive suite that addresses two critical needs for AI video generation: accessibility and speed.This all-in-one workflow provides two parallel generation pipelines:<ol><li>⚡ Lightning Fast (4-Step) Pipeline: Utilizes a specialized LoRA to generate videos in a fraction of the time, perfect for rapid prototyping and iteration.</li><li>🎨 High Fidelity (20-Step) Pipeline: The classic, high-quality generation process for when you demand the utmost visual fidelity from your outputs.</li></ol>Crucially, both versions are configured to run using GGUF-quantized models, dramatically reducing VRAM requirements and making this massive 14B parameter model accessible to users with consumer-grade hardware.<hr /><h3 id="key-features-and-highlights">✨ Key Features &amp; Highlights</h3><ul><li>Dual Mode Operation: Choose between speed and quality with two self-contained workflows in one JSON file. Easily enable/disable either section.</li><li>GGUF Quantization Support: Run the massive Wan 2.2 model without needing a professional GPU. Leverages <code>LoaderGGUF</code> and <code>ClipLoaderGGUF</code> nodes.</li><li>Extended Video Generation: The workflow includes built-in "Video S2V Extend" subgraphs. Each one adds 77 frames. The template is pre-configured with two extenders, resulting in a ~5-second video at 16 FPS. Want a longer video? Simply copy and paste more extender nodes!</li><li>Audio-Driven Animation: Faithfully implements the S2V model's core function: animating a reference image in sync with an uploaded audio file (e.g., music, speech).</li><li>Smart First-Frame Fix: Includes a clever hack to correct the first frame, which is often "overbaked" by the VAE decoder.</li><li>Detailed Documentation: The workflow itself is filled with informative notes and markdown nodes explaining crucial settings like batch size and chunk length.</li></ul><hr /><h3 id="how-it-works-(the-magic-behind-the-scenes)">🧩 How It Works (The Magic Behind the Scenes)</h3>The workflow is logically grouped into clear steps:<ol><li>Load Models (GGUF): The <code>LoaderGGUF</code> and <code>ClipLoaderGGUF</code> nodes load the quantized UMT5 text encoder and the main UNet model, drastically reducing VRAM load compared to full precision models.</li><li>Upload Inputs: You provide two key ingredients:<ul><li><code>ref_image</code>: The starting image you want to animate (e.g., a character portrait).</li><li><code>audio</code>: The sound file that will drive the motion and pacing of the animation.</li></ul></li><li>Encode Prompts &amp; Audio: Your positive and negative prompts are processed, and the audio file is encoded into a format the model understands using the Wav2Vec2 encoder.</li><li>Base Generation (<code>WanSoundImageToVideo</code>): The core node takes your image, audio, and prompts to generate the first latent video sequence.</li><li>Extend the Video (<code>Video S2V Extend</code> Subgraphs): This is where the length comes from. The latent output from the previous step is fed into a sampler (KSampler) alongside the audio context again to generate the next chunk of frames. These chunks are concatenated together.</li><li>Decode &amp; Compile: The final latent representation is decoded into images by the VAE, and the <code>CreateVideo</code> node stitches all the frames together with the original audio to produce your final MP4 file.</li></ol><hr /><h3 id="instructions-and-usage">⚙️ Instructions &amp; Usage</h3>Prerequisite: Download ModelsYou must download the following model files and place them in your ComfyUI <code>models</code> directory. The workflow includes handy markdown notes with direct download links.Essential Models:<ul><li><code>umt5-xxl-encoder-q4_k_m.gguf</code> → Place in <code>/models/clip/</code></li><li><code>Wan2.2-S2V-14B-Q5_0.gguf</code> → Place in <code>/models/unet/</code> (or <code>/models/diffusion/</code>)</li><li><code>wav2vec2_large_english_fp16.safetensors</code> → Place in <code>/models/audio_encoders/</code></li><li><code>wan_2.1_vae.safetensors</code> → Place in <code>/models/vae/</code></li></ul>For the 4-Step Lightning Pipeline:<ul><li><code>Wan2.2-Lightning_T2V-v1.1-A14B-4steps-lora_HIGH_fp16.safetensors</code> → Place in <code>/models/loras/</code></li></ul>Loading the Workflow<ol><li>Download the provided <code>video_wan2_2_14B_s2v.json</code> file.</li><li>In ComfyUI, drag and drop the JSON file into the window or use the Load button.</li></ol>Running the Workflow<ol><li>Upload Your Media:<ul><li>In the "LoadImage" node, upload your starting reference image.</li><li>In the "LoadAudio" node, upload your music or audio file.</li></ul></li><li>Enter Your Prompt:<ul><li>Modify the text in the "CLIP Text Encode (Positive Prompt)" node.</li><li>The negative prompt is already filled with a robust, standard negative.</li></ul></li><li>Choose Your Pipeline:<ul><li>To use the 4-Step Lightning pipeline (Fast): Ensure the <code>LoraLoaderModelOnly</code> node is correctly pointed to your Lightning LoRA file. The <code>Steps</code> primitive node for this section is already set to <code>4</code> and <code>CFG</code> to <code>1</code>.</li><li>To use the 20-Step pipeline (High Quality): The lower section of the workflow is already configured. The <code>Steps</code> are set to <code>20</code> and <code>CFG</code> to <code>6.0</code>. You can simply box-select the entire 20-step section and press <code>Ctrl+B</code> to disable the 4-step section if you wish to only run this one.</li></ul></li><li>Queue Prompt! Watch as your image comes to life, driven by your audio.</li></ol><hr /><h3 id="important-notes-and-tips">⚠️ Important Notes &amp; Tips</h3><ul><li>Batch Size Setting: The "Batch sizes" value (<code>3</code> by default) is not a traditional batch size. It must be set to <code>1 + [number of Video S2V Extend subgraphs]</code>. This workflow has 2 extenders, so the value is 3. If you add another extender, set it to 4.</li><li>Chunk Length: The default is <code>77</code> frames. This is a requirement of the model and should not be changed unless you know what you're doing.</li><li>Lightning LoRA Trade-off: The 4-step LoRA is incredibly fast but may result in a slight drop in coherence and quality compared to the 20-step generation. It's the perfect tool for finding the right seed and composition quickly.</li><li>GGUF vs. Safetensors: This workflow uses GGUF for the text and UNet models to save VRAM. You can replace the <code>LoaderGGUF</code> and <code>ClipLoaderGGUF</code> nodes with standard <code>UNETLoader</code> and <code>CLIPLoader</code> nodes if you have the VRAM to use the full <code>.safetensors</code> models, which may offer slightly better quality.</li></ul><hr /><h3 id="example-results">🎭 Example Results</h3>Prompt: "The man is playing the guitar. He looks down at his hands playing the guitar and sings affectionately and gently." Audio: A gentle acoustic guitar track.(You would embed a short video example generated by this workflow here)<hr /><h3 id="download-and-links">📁 Download &amp; Links</h3><ul><li>Download this Workflow JSON: <code>[Link to your uploaded JSON file]</code></li><li>Official Wan 2.2 Model Repo: <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.2_ComfyUI_Repackaged">HuggingFace - Comfy-Org/Wan_2.2_ComfyUI_Repackaged</a></li><li>Required GGUF Models: Search for <code>Wan2.2-S2V-14B-Q5_0.gguf</code> and <code>umt5-xxl-encoder-q4_k_m.gguf</code> on Hugging Face.</li></ul><hr /><h3 id="conclusion">💎 Conclusion</h3>This workflow demystifies the process of running the formidable Wan 2.2 S2V model. By integrating GGUF support and a dual-pipeline approach, it empowers users with limited hardware to experiment and create stunning, audio-synchronized animations. Whether you're quickly iterating with the Lightning LoRA or crafting a masterpiece with the full 20-step process, this suite has you covered.Happy generating! Feel free to leave a comment with your amazing creations or any questions.

Files with SHA256: 9fd2d9125aecec4ad3bb01546c14896ecd58308ece00d9e75bfcc1fc208fd296