Info

A simple script for detecting cuts in video footage and creating thumbnails from the detected scenes. Based on SCD script. In addition, a list of timecodes and markers is generated.

It's not an advanced and accurate AI tool, it simply detects differences in colour and pixel brightness. Therefore, it works reasonably well when the shots are quite bright and expressive. But it is reasonably fast and works in a browser. For simple cataloguing, so that you don't have to go through the whole film looking for a shot, it is enough.

The finished images can be downloaded as a zipped directory or as a collection of images in one file (imagewall).

If you need something more advanced, I recommend PySceneDetect , quite a good programme in my opinion.

A similar effect can be achieved with FFmpeg.
FFmpeg has quite a good filter for detecting differences between scenes. It works best for well-exposed shots, with old films it does not work very well. But this is unfortunately the problem with most of the publicly available tools.

A single line of code is enough:

ffmpeg -i <source file> -vf "select=gte(scene\,0.2),scale=200:-1,tile=10x50" -frames:v 1 -qscale:v 3 <destination pictures>.jpg

This is, of course, just an example. There could be many solutions.
- source and destination it is clear what it is.
- "gte(scene,0.2)" Detects a frame 20% changed from the previous frame.
- "tile=10x50" number of images per line and number of lines, in this example 10 horizontally and 50 vertically, for a total of 500 shots.
- "scale=200:-1" - thumbnails will have a horizontal size of 200px and a vertical size resulting from the aspect ratio of the image

If you have FFmpeg installed on your server or local computer, this is probably the easiest way to go, especially as you can basically use any type of video. And you can only use a few codecs on the website.