How to auto-detect scene boundaries before splitting
- Step 1Drop your video — Drop your video and pick a sensitivity threshold
- Step 2Set scene detection threshold — JAD runs select=scene+threshold and logs each cut
- Step 3Review timestamps or download split clips — Download the JSON list of scene timestamps
Frequently asked questions
How is the scene detection threshold set?+
JAD uses FFmpeg's scene detection filter which measures the average histogram difference between consecutive frames. A threshold of 0.4 (40%) detects major scene cuts while ignoring gradual brightness changes. Lower thresholds (0.2–0.3) detect softer transitions and fast-motion transitions. Higher thresholds (0.5–0.6) only detect very hard cuts. For typical interview or documentary content, 0.4 is a reliable starting point.
Can scene detection find fade-to-black transitions?+
FFmpeg's scene detection is primarily calibrated for hard cuts (abrupt histogram changes). Fade-to-black and cross-dissolve transitions involve gradual changes that may not exceed the threshold if set high. For fade detection, lower the threshold to 0.2–0.3, which will also produce more false positives on high-motion sequences. A post-processing pass filtering out very short clips can remove false positives.
What can I do with the detected scene timestamps?+
Scene timestamps can be used as chapter markers for YouTube uploads (paste the timestamp list into the video description). They can drive FFmpeg segment splitting for creating individual clip files. They can also feed into transcription pipelines (by aligning SRT subtitles to scene boundaries) or into editing software that accepts EDL (edit decision list) input.
Privacy first
All video processing runs locally in your browser using WebAssembly and FFmpeg. No file is ever uploaded — only metadata counters are saved for signed-in dashboard stats.