Video Clip Extractor for Lora Training

A utility tool to streamline the process of creating video clips and text captions for Lora training.

Features

• Load videos with high compatibility across formats
• Automatic clip detection
• Clip preview
• Frame adjustment with arrow keys
• Caption creation
• Export functionality with matching clip and text file names (1280x720 resolution)
• Resume capability for continued dataset building
• Custom output folder selection

Installation

Easy Installation (Recommended)

Simply run the run.bat file which will:

1. Create a virtual environment
2. Install all required dependencies
3. Launch the application

Manual Installation

If you prefer to install manually:

1. Create a virtual environment (recommended for compatibility):

   python -m venv venv
   venv\Scripts\activate
   

2. Install the required dependencies:

   pip install -r requirements.txt
   

3. Run the application:

   python main.py
   

Troubleshooting

• If you encounter issues with moviepy, the application will automatically try to use ffmpeg directly. Make sure ffmpeg is installed and available in your PATH.
• If you have compatibility issues, try using Python 3.9, 3.10, or 3.11 with the virtual environment.

Usage

1. Launch the application
2. Click "Choose Folder" to select where your clips will be saved
3. Click "Load Video" to open a video file
4. Click "Detect Scenes" to automatically find clips
5. Navigate between detected clips with "Previous Clip" and "Next Clip"
6. Use arrow keys to adjust clip boundaries
7. Preview the current clip
8. Enter a caption describing the clip
9. Click "Export Clip" to save both the clip and caption

Controls

• Arrow keys: Adjust clip boundaries
• Enter: Preview current clip
• Space: Play/Pause preview