PySceneDetect Documentation¶
PySceneDetect finds scene changes in video (fast cuts and fades) and can split the video, save
frames, or export the scene list. Use it as the scenedetect command, or as the
scenedetect Python package.
Install with pip install scenedetect (pip install scenedetect-headless on servers without
GUI libraries), or download the Windows build from
scenedetect.com/download. Splitting video requires
ffmpeg (or mkvmerge with split-video -m); the Windows build includes ffmpeg.
Quick Start¶
From the command line, find the scenes in video.mp4 and write one clip per scene:
scenedetect -i video.mp4 split-video
Add list-scenes for a CSV, save-images for still frames, or a detect-* command to
pick the detector (default: detect-adaptive). See the CLI reference.
The same from Python:
from scenedetect import detect, AdaptiveDetector, split_video_ffmpeg
scenes = detect("video.mp4", AdaptiveDetector())
split_video_ffmpeg("video.mp4", scenes)
scenes is a list of (start, end) timecode pairs. More Examples, each done both
ways, cover images, CSV output, detector tuning and more; see SceneManager for advanced workflows.
Reference¶
Most features exist on both sides. The Examples page shows each task done both ways, and every page links to its counterpart on the other side: each command to the function or class that does the same, and each detector to a table of its options, config file keys and constructor parameters.
CLI
Found a problem in the docs? Raise an issue on the PySceneDetect issue tracker. Development happens on Github at github.com/Breakthrough/PySceneDetect.
