LipSync in ComfyUI with ReActor and Wav2Lip - Make it Work

Learn to implement lip-sync functionality in ComfyUI through this comprehensive 29-minute tutorial that covers local installation and troubleshooting of Wave2Lip and ReActor nodes. Master the setup process for ComfyUI 0.3.66 portable version with Python 3.13 embedded, including proper installation of dependencies like FFmpeg and Git. Explore built-in templates such as Humo/HERMA and VAN 2.2 for audio-driven video generation, understanding their capabilities and limitations with different character types. Discover external workflows that extend beyond default templates, comparing version trade-offs between ComfyUI 0.2.3 and newer releases. Navigate common installation challenges including ReActor errors, Python version compatibility issues, and audio library conflicts with detailed troubleshooting solutions. Follow step-by-step instructions for cloning custom nodes, installing requirements through embedded Python, and resolving missing node errors through ComfyUI-Manager. Address specific technical issues like soxr/librosa import problems using sitecustomize workarounds and proper wheel installations. Execute complete end-to-end pipeline walkthroughs from audio input to final lip-synced video output, with practical tips for optimizing clip length, stability, head motion, and frame rates for professional results.

Syllabus

0:00 Intro — Lip sync in ComfyUI on a local PC
0:22 Resources, links, and what we’ll cover
0:50 Installing ComfyUI 0.3.66 portable and Python 3.13
1:10 Using ComfyUI-Manager to install missing nodes
1:30 Built-in templates overview Humo/HERMA & VAN 2.2
1:44 Humo Video Generation test — setup, prompts, limits
2:52 Non-human characters issue Muppet example, no lip sync
3:46 VAN 2.2 Audio-Driven Video — workflow and results
4:45 Extending clip length and stability tips
5:04 External workflows: Wave2Lip options not in templates
5:16 Comfy-Wave2Lip + React; Lip Sync Warper Python 3.11 note
6:02 Version trade-offs: ComfyUI 0.2.3 vs latest features
6:23 If React install is painful — quick online alternative demo
7:45 Domo AI “Talking Avatar” demo: TTS, upload, render steps
9:37 Back to local: Wave2Lip pipeline, FFmpeg install & PATH setup
11:01 Install Git; clone custom nodes into custom_nodes
14:11 Install requirements.txt with embedded Python/pip
17:55 Load sample Wave2Lip workflow; download model to checkpoints
19:01 Fix missing nodes via Manager; common React errors explained
20:54 React troubleshooting: correct Python version & prebuilt wheels
22:06 Commands to install the right wheel; re-run requirements
24:31 Fixing soxr/librosa import issues sitecustomize workaround
26:27 Relaunch ComfyUI; end-to-end pipeline walkthrough
27:34 Results: lip-sync quality, head motion, FPS/length tips
28:16 Wrap-up — files on Patreon, Q&A, next steps