Scenecut Extractor

Extract scenecuts from video files using ffmpeg.

This tool uses the select filter from ffmpeg to determine the scene cut probability of adjacent frames, and allows users to determine which frames (or at which timestamps) the scene cuts happen.

Note: Previous versions installed a scenecut_extractor executable. To harmonize it with other tools, now the executable is called scenecut-extractor. Please ensure you remove the old executable (e.g. run which scenecut_extractor and remove the file).

Author: Werner Robitza werner.robitza@gmail.com

Contents:

• Requirements
• Installation
• Usage
• Extended Usage
• API
• Alternatives and Related Tools
• Contributors
• License

Requirements

• Python 3.10 or higher
• FFmpeg:
  • download a static build from their website)
  • put the ffmpeg executable in your $PATH

Installation

Using uv:

uvx scenecut-extractor

Using pipx:

pipx install scenecut-extractor

Or, using pip:

pip3 install --user scenecut-extractor

Usage

Run:

scenecut-extractor <input-file>

This might take a while depending on the length of your input file, and then output a list of scene cuts in JSON format:

[
  {
    "frame": 114,
    "pts": 114.0,
    "pts_time": 3.8,
    "score": 0.445904
  },
  {
    "frame": 159,
    "pts": 159.0,
    "pts_time": 5.3,
    "score": 0.440126
  }
]

To extract the scene cuts, use the -x flag and optionally specify an output directory with -d:

scenecut-extractor <input-file> -x -d output-directory

This will create a directory called output-directory and put the extracted scenes in there. The filenames will be the same as the input file, but with the scene times appended to them.

Note: Cutting may not be frame-accurate. To be precise, you have to re-encode the video. Use the --no-copy flag to do this. The output will use libx264 encoding with CRF 23 and the medium preset by default, which gives a good balance between quality and file size. Use --crf (0–51, lower means higher quality) and --preset to change this, for example:

scenecut-extractor <input-file> -x --no-copy --crf 18 --preset slow

Extended Usage

The command supports the following arguments and options, see scenecut-extractor -h:

usage: scenecut-extractor [-h] [-t THRESHOLD] [-o {all,frames,seconds}]
                          [-of {json,csv}] [-x] [-d OUTPUT_DIRECTORY]
                          [--no-copy] [--crf CRF]
                          [--preset {ultrafast,superfast,veryfast,faster,fast,medium,slow,slower,veryslow,placebo}]
                          [-e OUTPUT_EXTENSION] [-p] [-v] [-O OUTPUT_FILE]
                          [--ffmpeg-path FFMPEG_PATH]
                          input

scenecut_extractor v0.9.2

positional arguments:
  input                 input file

options:
  -h, --help            show this help message and exit
  -t, --threshold THRESHOLD
                        threshold (between 0 and 1) (default: 0.3)
  -o, --output {all,frames,seconds}
                        output which information (default: all)
  -of, --output-format {json,csv}
                        output in which format (default: json)
  -x, --extract         extract the scene cuts (default: False)
  -d, --output-directory OUTPUT_DIRECTORY
                        Set the output directory. Default is the current
                        working directory. (default: None)
  --no-copy             Don't stream-copy, but re-encode the video. (default:
                        False)
  --crf CRF             CRF value (0-51) for libx264. Only used when re-
                        encoding (--no-copy). (default: 23)
  --preset {ultrafast,superfast,veryfast,faster,fast,medium,slow,slower,veryslow,placebo}
                        libx264 preset. Only used when re-encoding (--no-
                        copy). (default: medium)
  -e, --output-extension OUTPUT_EXTENSION
                        Override the output file extension for extracted
                        scenes (e.g., '.mkv'). Default: same as input.
                        (default: None)
  -p, --progress        Show a progress bar on stderr (default: False)
  -v, --verbose         Print verbose info to stderr (default: False)
  -O, --output-file OUTPUT_FILE
                        Write output to a file instead of stdout (default:
                        None)
  --ffmpeg-path FFMPEG_PATH
                        Path to ffmpeg executable (default: ffmpeg)

You can use the -t parameter to set the threshold that ffmpeg internally uses (between 0 and 1) – if you set it to 0, all frames will be printed with their probabilities.

API

This program has a simple API that can be used to integrate it into other Python programs.

For more information see the API documentation.

Alternatives and Related Tools

For extended scene detection features such as automatic splitting or perceptual hashing, you may want to check out PySceneDetect.

If you are interested in generating chapters automatically from various media files, have a look at chaptertool (which also uses this library).

Contributors

xqrdot 🤔

License

scenecut_extractor, Copyright (c) 2018–2026 Werner Robitza

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.