<h1 id="wanvideo-comfyui-runpod-setup-guide">🎥 WanVideo ComfyUI RunPod Setup Guide</h1>This comprehensive guide will walk you through setting up and using the WanVideo ComfyUI environment on RunPod for AI video generation. Wan needs alot or vram to get outputs ata resonable speed. 48GB for $0.44 an hour is a pretty good deal IMO Nothing to download from here. This is a Runpod Template with all the models and workflows added. WAN 2.1 Video - ComfyUI Full - T2.0 - Running on CUDA 2.5 <a target="_blank" rel="ugc" href="https://runpod.io/console/deploy?template=6k2saccgx8&amp;ref=0eayrc3z">https://runpod.io/console/deploy?template=6k2saccgx8&amp;ref=0eayrc3z</a> <pre><code>## UPDATE
20/03/25
Notice a Pytorch Bug on using Community Cloud GPUs. 
EDIT: Its with just the 5090 cards with the blackwell architecture

19/03/25
'Error: 'NoneType' object is not callable fix'
I added a depth 0 to ComfyUI and implemented nodes to reduce the container size, but these changes introduced several bugs. I've since removed them, and everything should now work much better.

17/03/25
- Added environment variables to control setup behavior:
 - `SKIP_DOWNLOADS=true`: Skip downloading models
 - `SKIP_NODES=true`: Skip verification custom nodes (nodes are packaged into the container for faster build)</code></pre><h2 id="getting-started">🚀 Getting Started</h2><h3 id="step-1:-deploying-your-pod">Step 1: Deploying Your Pod</h3><ol><li>Sign up/login to <a target="_blank" rel="ugc" href="https://runpod.io/?ref=0eayrc3z">RunPod</a></li><li>Navigate to "Deploy" → "Template"</li><li>Search for "WAN 2.1 Video - ComfyUI Full - T1.0" template</li><li>Select the hardware:<ul><li>Recommended GPU: RTX A40 (minimum 48GB VRAM)</li><li>Storage: 60GB minimum (100GB recommended)</li><li>Filter GPU's above CUDA 2.4</li></ul></li><li>Click "Deploy" to launch your pod</li></ol><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/1d70d392-80aa-4e33-8fef-c6b9e6f6f780/width=525/1d70d392-80aa-4e33-8fef-c6b9e6f6f780.jpeg" />Step 2: Initial SetupOnce deployed, your pod will automatically:<ul><li>Download all required models (takes ~10 minutes)</li><li>Install custom nodes</li><li>Set up the environment</li></ul>You'll see the following message when setup is complete:<pre><code>⭐⭐⭐⭐⭐ ALL DONE - STARTING COMFYUI ⭐⭐⭐⭐⭐
</code></pre><h3 id="step-3:-accessing-your-environment"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/774aa9e4-b8b1-47ca-80cf-6a657c980bd6/width=525/774aa9e4-b8b1-47ca-80cf-6a657c980bd6.jpeg" /> <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/d5193284-fc54-43dc-860c-4385402c0be1/width=525/d5193284-fc54-43dc-860c-4385402c0be1.jpeg" /> Step 3: Accessing Your Environment</h3>From your pod's detail page, access:<ol><li>ComfyUI Interface:<ul><li>Port 8188 (primary interface for creating videos)</li><li>Wait for this to turn green after setup completes</li><li><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/c7e1c67b-6672-4bc7-b99d-79cf75b99d05/width=525/c7e1c67b-6672-4bc7-b99d-79cf75b99d05.jpeg" /></li></ul></li><li>JupyterLab:<ul><li>Port 8888 (available immediately, even during setup)</li><li>Use for file management, terminal access, and notebook interactions</li><li><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/bb9acfc9-2be5-4bc2-831c-6cbbc2c571d0/width=525/bb9acfc9-2be5-4bc2-831c-6cbbc2c571d0.jpeg" /></li></ul></li><li>Image Browser:<ul><li>Port 8181 (for managing your output files)</li><li>View and organize generated videos and images</li><li><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/ec95fa92-044d-41e0-b623-2c11d3bcbcd1/width=525/ec95fa92-044d-41e0-b623-2c11d3bcbcd1.jpeg" /> </li></ul></li></ol><h2 id="managing-models">💾 Managing Models</h2><h3 id="downloading-additional-models">Downloading Additional Models</h3>This will only download everything you need for Wan_Image2Video_720pAFIX.json workflow. If you want to use the other workflows run the ./<a target="_blank" rel="ugc" href="http://download-files.sh">download-files.sh</a> from terminal and it will download all the models for Kajai's workflow.Use the flexible model download system:<ol><li>Edit the configuration file:<pre><code>files.txt
</code></pre><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/532d3334-a9de-4ed7-b224-27483bfae33c/width=525/532d3334-a9de-4ed7-b224-27483bfae33c.jpeg" /></li><li>Add model entries using this format:<pre><code>type|folder|filename|url
</code></pre>Examples:<pre><code>normal|checkpoints|realistic_model.safetensors|https://huggingface.co/org/model/resolve/main/model.safetensors

gdrive|loras|animation_style.safetensors|https://drive.google.com/uc?id=your_file_id
</code></pre></li><li>Run the download script:<pre><code>./download-files.sh
</code></pre><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/df3e79de-de5f-4a8f-9689-e692b2d73da7/width=525/df3e79de-de5f-4a8f-9689-e692b2d73da7.jpeg" /></li></ol>WAN 2.1 Models<a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/tree/main">https://huggingface.co/Kijai/WanVideo_comfy/tree/main</a><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/tree/main/split_files">https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/tree/main/split_files</a><h3 id="pre-installed-models">Pre-installed Models</h3>The template comes with these key models:<ul><li>Wan 2.1 Models:<ul><li><code>Wan2_1-I2V-14B-720P_fp8_e4m3fn.safetensors</code> (Base video model)</li><li><code>wan_2.1_vae.safetensors</code> (VAE)</li></ul></li><li>Text Encoders:<ul><li><code>umt5_xxl_fp16.safetensors</code> (Advanced text encoder)</li></ul></li><li>CLIP Vision:<ul><li><code>clip_vision_h.safetensors</code> (Enhanced vision model)</li></ul></li></ul><h2 id="using-comfyui-for-video-generation">🎨 Using ComfyUI for Video Generation</h2><h3 id="step-1:-load-a-workflow">Step 1: Load a Workflow</h3><ol><li>Access ComfyUI interface (port 8188)</li><li>Click on the folder icon in the top menu</li><li>Navigate to the default workflows folder</li><li>Select one of the pre-configured workflows</li></ol><h3 id="step-2:-customize-your-generation">Step 2: Customize Your Generation</h3><ol><li>Modify text prompts to describe your desired video</li><li>Adjust settings:<ul><li>CFG Scale: 7-9 recommended for quality (higher = more prompt adherence)</li><li>Steps: 25+ for better quality (more steps = more refinement)</li><li>Resolution: Start with 512x512 for tests, increase for final outputs</li><li>Frame count: Determines video length</li></ul></li></ol><h3 id="step-3:-generate-and-view-results">Step 3: Generate and View Results</h3><ol><li>Click "Queue Prompt" to start generation</li><li>Monitor progress in the ComfyUI interface</li><li>When complete, view your video in the output panel</li><li>Access all outputs via the Image Browser (port 8181)</li></ol><h2 id="managing-your-files">📊 Managing Your Files</h2><h3 id="using-jupyterlab">Using JupyterLab</h3><ol><li>Access JupyterLab (port 8888)</li><li>The workspace folder contains:<ul><li><code>/ComfyUI</code> - Main application and models</li><li>Files for downloading additional models</li><li>Notebook for image/video browsing</li></ul></li></ol><h3 id="using-image-browser">Using Image Browser</h3><ol><li>Access the browser interface (port 8181)</li><li>Browse your generated content by:<ul><li>Creation date</li><li>Filename</li><li>Metadata</li></ul></li><li>Right-click on items for additional options (download, delete, etc.)</li></ol><h2 id="advanced-features">🔧 Advanced Features</h2><h3 id="ssh-access">SSH Access</h3>To enable SSH:<ol><li>Set your <code>PUBLIC_KEY</code> in the template settings before deployment</li><li>Connect using the command shown in your pod's connect options</li></ol><h3 id="custom-nodes">Custom Nodes</h3>The template includes these pre-installed node collections:<ul><li>Workflow utilities (<code>cg-use-everywhere</code>, <code>ComfyUI-Manager</code>)</li><li>UI enhancements (<code>rgthree-comfy</code>, <code>was-node-suite-comfyui</code>)</li><li>Video-specific nodes (<code>ComfyUI-WanVideoWrapper</code>, <code>ComfyUI-VideoHelperSuite</code>)</li><li>Performance optimizers (<code>ComfyUI-Impact-Pack</code>)</li></ul><h2 id="troubleshooting">🛠️ Troubleshooting</h2>If you encounter issues:<ol><li>ComfyUI not starting:<ul><li>Check JupyterLab terminal for logs</li><li>Ensure models downloaded correctly</li></ul></li><li>Models not loading:<ul><li>Verify files exist in <code>/ComfyUI/models/</code> directories</li><li>Check file sizes to ensure complete downloads</li></ul></li><li>Custom node problems:<ul><li>Try reinstalling via ComfyUI Manager</li><li>Restart your pod if necessary</li></ul></li></ol><h2 id="tips-for-best-results">🎯 Tips for Best Results</h2><ul><li>Use detailed prompts with specific descriptions</li><li>Increase CFG and step count for higher quality videos</li><li>Save your successful workflows for future use</li><li>Monitor VRAM usage and adjust resolution accordingly</li><li>Use the Image Browser to organize and review your outputs</li></ul>Need more help? Check the <a target="_blank" rel="ugc" href="http://readme.md">readme.md</a> file in JupyterLab or reach out to the RunPod community!

Runpod WAN 2.1 Img2Video Template - ComfyUI

<ol><li>ComfyUI Interface:<ul><li>Port 8188 (primary interface for creating videos)</li><li>Wait for this to turn green after setup completes</li></ul></li><li>JupyterLab:<ul><li>Port 8888 (available immediately, even during setup)</li><li>Use for file management, terminal access, and notebook interactions</li></ul></li><li>Image Browser:<ul><li>Port 8181 (for managing your output files)</li><li>View and organize generated videos and images</li></ul></li></ol>

Files with SHA256: 5aef4bedf5a5e27bf3ce810c60fac7c32f059fb9a6627a3793e684e003f2d270