As we discussed in #128, there seems to be some merit to moving the -m/--min-scene-len option, which is duplicated between detect-content and detect-threshold to the main option parsing. My interest here isn't just in reducing the duplication, though that's certainly a nice side effect, but in adding the option to drop short scenes instead of having them combined with neighbors, which seems to be the current behavior.
So here's what I'm thinking of doing after #128 is merged:
- Remove
--min-scene-len from detect-content and detect-threshold,
- Add it to
scenedetect_cli.
- Add a new option, something like
--drop-short-scenes, which, when True, has detect-content and detect-threshold ignore the --min-scene-len option, deferring handling of the option until after the scene list is generated.
- Update
context.py to drop scenes shorter than --min-scene-len when --drop-short-scenes is True.
Thoughts?
As we discussed in #128, there seems to be some merit to moving the
-m/--min-scene-lenoption, which is duplicated betweendetect-contentanddetect-thresholdto the main option parsing. My interest here isn't just in reducing the duplication, though that's certainly a nice side effect, but in adding the option to drop short scenes instead of having them combined with neighbors, which seems to be the current behavior.So here's what I'm thinking of doing after #128 is merged:
--min-scene-lenfromdetect-contentanddetect-threshold,scenedetect_cli.--drop-short-scenes, which, whenTrue, hasdetect-contentanddetect-thresholdignore the--min-scene-lenoption, deferring handling of the option until after the scene list is generated.context.pyto drop scenes shorter than--min-scene-lenwhen--drop-short-scenesisTrue.Thoughts?