🎭 Ultimate Face Swap Video Workflow for ComfyUI
Swap any face onto any AI-generated video seamlessly using
ReActor + VideoHelperSuite pipeline.
✅ FEATURES:
- High-quality face swap on video frames
- CodeFormer face restoration built-in
- RetinaFace detection for accuracy
- GIF/Video output ready
- Audio preservation support
- Simple 3-node pipeline
⚙️ REQUIRED NODES:
- comfyui-reactor
- comfyui-videohelpersuite
📦 MODELS NEEDED:
- inswapper_128.onnx
- codeformer-v0.1.0.pth
- retinaface_resnet50
🚀 HOW TO USE:
1. Load your source video
2. Load your target face image
3. Hit Run — done!
💡 TIPS:
- Use 512px width for faster processing
- codeformer_weight 0.5 = best balance
- face_restore_visibility 1.0 for clean output
⭐ If this helped you, leave a review!
Description
🚀 v1.0 - First Release
WHAT'S NEW:
- Initial workflow release
- ReActor Fast Face Swap integration
- VideoHelperSuite pipeline setup
- CodeFormer restoration support
TESTED ON:
- ComfyUI latest version
- inswapper_128.onnx
- codeformer-v0.1.0.pth
SETTINGS USED:
- Resolution: 512px
- Face restore visibility: 1.0
- CodeFormer weight: 0.50
- Frame rate: 16fps
- Output: GIF/AnimateDiff
KNOWN LIMITATIONS:
- Best results with front-facing images
- Requires clear source face image
COMING IN v1.1:
- Multi-face support
- Higher resolution option
- Batch processing
FAQ
Comments (3)
Worked surprisingly well tbh, but got an OOM in longer vids (15 secs)
I am stuck with the comfy not recognizing the reactor node, shows its missing but when i go to install it it gives me an error that its already there?
Delete the node and reinstall it again.