ComfyUI Video Save Node

A custom ComfyUI node that saves video outputs as H.264 .mp4 files with unique naming and returns the file path as a string.

Features

• 🎥 Saves video tensors as H.264 encoded .mp4 files
• 🔢 Generates unique filenames with timestamp and UUID
• 📁 Configurable output directory
• ⚙️ Adjustable quality settings (CRF 0-51)
• 🎬 Configurable frame rate
• 📝 Returns full file path as string output
• 🔄 Automatic tensor format detection and conversion
• ⚡ Optional FFmpeg re-encoding for better compression

Installation

1. Copy the comfyui_video_save_node folder to your ComfyUI custom_nodes directory:

   cp -r comfyui_video_save_node /path/to/ComfyUI/custom_nodes/

2. Install the required Python dependencies:

   cd /path/to/ComfyUI/custom_nodes/comfyui_video_save_node
   pip install -r requirements.txt

3. (Optional but recommended) Install FFmpeg for better H.264 encoding:

   • macOS: brew install ffmpeg
   • Ubuntu: sudo apt install ffmpeg
   • Windows: Download from FFmpeg website

4. Restart ComfyUI

Usage

1. Add the "Save Video (H.264)" node to your workflow
2. Connect a video output from another node (e.g., VEO3 API node) to the video input
3. Configure the parameters:
   • filename_prefix: Prefix for the output filename (default: "video_output")
   • fps: Frame rate for the output video (default: 30)
   • quality: H.264 CRF quality, lower = better quality (default: 23)
   • output_dir: Optional custom output directory (uses ComfyUI output folder by default)
4. The node will output the full file path as a string

Output Format

The node generates filenames in the format:

{filename_prefix}_{timestamp}_{unique_id}.mp4

Example: video_output_20241213_143022_a1b2c3d4.mp4

Node Parameters

Parameter	Type	Default	Description
video	VIDEO	-	Input video tensor from previous nodes
filename_prefix	STRING	"video_output"	Prefix for the output filename
fps	INT	30	Frame rate (1-120)
quality	INT	23	H.264 CRF quality (0-51, lower = better)
output_dir	STRING	""	Optional custom output directory

Output

Output	Type	Description
file_path	STRING	Full path to the saved video file

Technical Details

• The node automatically detects and converts various video tensor formats
• Supports both batch and single video tensors
• Normalizes pixel values to 0-255 range
• Handles RGB/BGR conversion for OpenCV compatibility
• Uses OpenCV for initial encoding, with optional FFmpeg re-encoding
• Generates unique filenames to prevent overwrites

Troubleshooting

• "Invalid or empty video tensor": Ensure the input video tensor is valid and not empty
• FFmpeg errors: FFmpeg is optional; the node will fall back to OpenCV encoding
• File permission errors: Ensure ComfyUI has write permissions to the output directory

License

This project is open source. Feel free to modify and distribute.