chunk_encoder/README.md

Chunk Encoder

Note: This project is a work in progress. The scripts, their features, and command-line arguments are subject to change.

A collection of Python scripts to automate a video encoding workflow. The process involves detecting scenes, cutting the source video into smaller segments, encoding each segment individually, and muxing them back into a final file.

Prerequisites

Ensure the following command-line tools are installed and available in your system's PATH:

• ffmpeg
• ffprobe
• SvtAv1EncApp.exe (from the SVT-AV1 project)
• mkvmerge (from the MKVToolNix package)

Workflow

The scripts are designed to be run in sequence. They create and use local directories (cuts, segments, temp) to store intermediate files.

1. (Optional) Detect Crop

If your video has black bars, you can use cropdetect.py to find the correct crop values.

python cropdetect.py "path/to/your/video.mkv"

2. Cut Video into Scenes

Use scene_cutter.py to analyze the video, detect scene changes, and split the source into lossless segments in the cuts/ directory.

python scene_cutter.py "path/to/your/video.mkv"

3. Encode Segments

Choose one of the encoder scripts to process the files from the cuts/ directory. Encoded files will be placed in the segments/ directory.

Option A: VMAF-based Encoding (Recommended) Use vmaf_encoder.py to encode each segment to a target VMAF quality level.

python vmaf_encoder.py --target-vmaf 96.0

Option B: Static CRF Encoding Use static_encoder.py to encode all segments with a single, fixed CRF value.

python static_encoder.py --crf 27

4. Mux Final Video

Use segment_muxer.py to combine all the encoded .mkv files from the segments/ directory into a single output file.

python segment_muxer.py --cleanup

Using the --cleanup flag will automatically remove the cuts/ and segments/ directories