<h2 id="ultimate-ai-video-production-workflow-(comfyui)">🚀 <strong>Ultimate AI Video Production Workflow (ComfyUI)</strong></h2><p>This is a <strong>complete end-to-end AI video pipeline</strong> that combines:</p><h3 id="qwen-edit-2509-image-editing">🧩 <strong>Qwen-Edit 2509</strong> → Image Editing</h3><h3 id="wan-animate-2.2-character-animation">🎭 <strong>Wan Animate 2.2</strong> → Character Animation</h3><h3 id="seedvr2-upscaler-final-4k-enhancement">📺 <strong>SeedVR2 Upscaler</strong> → Final 4K Enhancement</h3><p>Everything is packed into a <strong>clean, modular, subgraphed workflow</strong> that removes the headache of giant node spaghetti.</p><p>This workflow allows you to:</p><ul><li><p>Edit any reference image</p></li><li><p>Animate the edited character using a real motion video</p></li><li><p>Upscale and clean the final output</p></li><li><p>Produce 4K smooth AI videos</p></li><li><p>Run everything inside ComfyUI with optimized VRAM usage</p></li></ul><p>Perfect for:</p><ul><li><p>AI character reels</p></li><li><p>Cosplay transformations</p></li><li><p>Cinematic AI edits</p></li><li><p>TikTok / IG Reels</p></li><li><p>YouTube shorts</p></li><li><p>Film-style previews</p></li></ul><hr /><h1 id="features">🎛️ <strong>Features</strong></h1><h3 id="1.-qwen-edit-2509-fp8-(image-editing-engine)">✨ <strong>1. Qwen-Edit 2509 FP8 (Image Editing Engine)</strong></h3><ul><li><p>Outfit changes</p></li><li><p>Face cleanup / enhancement</p></li><li><p>Makeup edits</p></li><li><p>Background removal</p></li><li><p>Style injection</p></li><li><p>Very VRAM-friendly</p></li></ul><h3 id="2.-wan-animate-2.2-fp8-(character-animation)">🎬 <strong>2. Wan Animate 2.2 FP8 (Character Animation)</strong></h3><ul><li><p>Robust identity preservation</p></li><li><p>Smooth motion from reference video</p></li><li><p>Pose &amp; face control</p></li><li><p>Windowed batching for long videos</p></li><li><p>Subgraphed for easy navigation</p></li></ul><h3 id="3.-seedvr2-upscaler-(video-restore-+-4k)">📈 <strong>3. SeedVR2 Upscaler (Video Restore + 4K)</strong></h3><ul><li><p>2× / 4× upscaling</p></li><li><p>Face enhance option</p></li><li><p>Removes noise and artifacts</p></li><li><p>Makes AI videos look real and cinematic</p></li></ul><h3 id="bonus-modules">🔧 <strong>Bonus Modules</strong></h3><ul><li><p>VRAM Cleaner</p></li><li><p>Resolution Selector</p></li><li><p>Reference Routing System</p></li><li><p>Organized Subgraphs</p></li><li><p>Cleaner wiring = easier editing</p></li></ul><hr /><h1 id="how-to-use-this-workflow">🧰 <strong>How to Use This Workflow</strong></h1><h3 id="1-load-your-reference-image">1️⃣ Load your <strong>reference image</strong></h3><p>Use Qwen-Edit to perform any image cleanup or enhancement before animation.</p><h3 id="2-load-your-motion-video">2️⃣ Load your <strong>motion video</strong></h3><p>Wan Animate will replicate the motion on your edited character.</p><h3 id="3-adjust-resolution-and-pre-process-settings">3️⃣ Adjust resolution &amp; pre-process settings</h3><p>Resolution presets included (320 → 1280).</p><h3 id="4-run-wan-animate-2.2">4️⃣ Run Wan Animate 2.2</h3><p>Generates stable character animation frames.</p><h3 id="5-pass-output-into-seedvr2">5️⃣ Pass output into SeedVR2</h3><p>Upscale to 2K / 4K depending on hardware.</p><h3 id="6-save-final-video">6️⃣ Save final video</h3><p>Both preview &amp; HQ versions included.</p><p></p><h3 id="qwen-image-edit-fp8-(diffusion-model-text-encoder-and-vae)">🖼️ Qwen Image Edit FP8 (Diffusion Model, Text Encoder, and VAE)</h3><p></p><p>These are hosted on the <strong>Comfy-Org</strong> Hugging Face page.</p><ul><li><p><strong>Diffusion Model (qwen_image_edit_fp8_e4m3fn.safetensors):</strong> <code>https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI/blob/main/split_files/diffusion_models/qwen_image_edit_fp8_e4m3fn.safetensors</code></p></li><li><p><strong>Text Encoder (qwen_2.5_vl_7b_fp8_scaled.safetensors):</strong> <code>https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI/blob/main/split_files/text_encoders/qwen_2.5_vl_7b_fp8_scaled.safetensors</code></p></li><li><p><strong>VAE (qwen_image_vae.safetensors):</strong> <code>https://huggingface.co/Comfy-Org/Qwen-Image-Edit_ComfyUI/blob/main/split_files/vae/qwen_image_vae.safetensors</code></p></li></ul><hr /><p></p><h3 id="wan-2.2-animate-14b-fp8-(diffusion-model-text-encoder-and-vae)">💃 Wan 2.2 Animate 14B FP8 (Diffusion Model, Text Encoder, and VAE)</h3><p></p><p>The components are spread across related community repositories.</p><ul><li><p><strong>Diffusion Model (Wan2_2-Animate-14B_fp8_e4m3fn_scaled_KJ.safetensors):</strong> <code>https://huggingface.co/Kijai/WanVideo_comfy_fp8_scaled/blob/main/Wan22Animate/Wan2_2-Animate-14B_fp8_e4m3fn_scaled_KJ.safetensors</code></p></li><li><p><strong>Text Encoder (umt5_xxl_fp8_e4m3fn_scaled.safetensors):</strong> <code>https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors</code></p></li><li><p><strong>VAE (wan2.2_vae.safetensors):</strong> <code>https://huggingface.co/Comfy-Org/Wan_2.2_ComfyUI_Repackaged/blob/main/split_files/vae/wan2.2_vae.safetensors</code></p></li><li><p><code>Lightx2v : https://huggingface.co/Kijai/WanVideo_comfy/tree/main/Lightx2v</code></p></li></ul><hr /><p></p><h3 id="seedvr2-diffusion-model-(fp8)">💾 SeedVR2 Diffusion Model (FP8)</h3><p></p><ul><li><p><strong>Diffusion Model (seedvr2_ema_3b_fp8_e4m3fn.safetensors):</strong> <code>https://huggingface.co/numz/SeedVR2_comfyUI/blob/main/seedvr2_ema_3b_fp8_e4m3fn.safetensors</code></p></li></ul><hr /><h1 id="why-this-workflow-is-easy-to-use">🧱 <strong>Why This Workflow is Easy to Use</strong></h1><p>I wrapped the heavy sections inside <strong>subgraphs</strong>, so you don’t have to scroll through 200 tangled nodes.</p><p>Subgraphs included:</p><ul><li><p>Wan Animate Engine</p></li><li><p>Qwen-Edit Module</p></li><li><p>SeedVR2 Upscaler</p></li><li><p>VRAM Cleaner</p></li><li><p>Resolution Manager</p></li></ul><p>Makes editing, debugging, and expanding the workflow extremely clean and fast.</p><hr /><h1 id="credits">❤️ <strong>Credits</strong></h1><ul><li><p>Kijai for FP8 Wan Animate</p></li><li><p>Comfy-Org for Qwen-Edit</p></li><li><p>ByteDance for SeedVR2</p></li><li><p>Community testers &amp; contributors</p></li></ul><p></p>

Files with SHA256: 4bfa9a780e7055378951391875d94b058a5cbffc37adef657e563ab8f70d82a2