HDHomeRun H 264?

Thanks for the logs. So it looks like Comskip isn’t being aggressive enough to detect the ads, it’s finding something but it’s also confusing some of the ads for the main show

2025-07-14T13:05:09 MCEBuddy.CommercialScan.Scanner → ParseEDL: Cut Segment Start:1855.690 End:1859.960 Action:0
→ ConvertCutListToKeepList: Keep segment Start:0.000 Stop:1855.690
→ ConvertCutListToKeepList: Keep end segment Start:1859.960 Stop:1860.993
→ ConvertKeepListToCutList: Skip segment, too small Start:0.000 Stop:0.000
→ ConvertKeepListToCutList: Cut segment Start:1855.690 Stop:1859.960
→ ConvertKeepListToCutList: Skip end segment, too small Start:1860.993 Stop:1860.993
→ ConvertKeepListToCutList Done

I would recommend trying to start with making Comskip more aggressive..

A really neat way is to use AI to help make it more aggressive and use a custom comskip ini (use the included Comskip INI editor to make changes to the INI and understand the changes you’re making). You can use the custom BOT we have on this forum.

image

Just copy paste the contents of your INI file and tell the AI BOT Helper to make it more aggressive. I just did that and here are some suggestions.

To make your Comskip INI file more aggressive in detecting commercials, you can adjust several parameters. Remember that increasing aggressiveness can lead to false positives (cutting parts of the show), so proceed cautiously and test after each change. Start with small adjustments and observe the results.

Here’s a breakdown of settings and suggestions for increasing aggressiveness:

High Impact Settings (Adjust with Caution):

  • detect_method (Currently 127): This is a bitmask controlling which detection methods are used. The value 127 enables almost all methods. You could try experimenting with slightly different combinations if you find certain methods are more prone to false positives. A good starting point for more aggressive detection is to enable every method.
  • min_commercial_size (Currently 4): This sets the minimum length (in seconds) of a detected commercial segment. Lowering this value makes Comskip more sensitive to shorter commercials, but it also increases the risk of false positives. Try reducing it gradually (e.g., to 3 or even 2) and observe results.
  • logo_threshold (Currently 0.75): This controls how similar a detected logo must be to a known logo to be considered a match. Lowering this value makes it more likely to detect logos even with slight variations, increasing aggressiveness. Try reducing to 0.7, 0.65, or even lower (but below 0.5 is extremely aggressive).
  • logo_max_percentage (Currently 0.18): This is the maximum percentage of the frame that can be covered by a logo to be detected. Decreasing this value will make logo detection more sensitive and aggressive but will potentially increase false positives.
  • logo_percentage_threshold (Currently 0.92): This threshold determines how much of a logo needs to be present for Comskip to be sure of a detection. Lowering it will make it more sensitive to partial logo detections and make it more aggressive.
  • max_commercialbreak (Currently 600): This sets the maximum length (in seconds) of a commercial break. Increasing this value slightly might help capture longer commercial breaks. But it will likely not make a huge difference.
  • silence_threshold (Currently 100): Lowering this increases sensitivity to silence. Experiment with reducing this value to detect quieter commercials but be aware that background sounds could trigger false positives.
  • max_silence_blocks (Currently 40): Adjusts how many consecutive blocks of silence are needed to identify a commercial break. Lower values make it more aggressive.

Medium Impact Settings:

  • non_uniformity (Currently 500): Lower values increase sensitivity to non-uniformity in brightness, but be aware of false positives from scene changes.
  • max_brightness, max_avg_brightness, test_brightness: Experiment with lowering these values to increase sensitivity to brightness changes commonly associated with commercials.
  • punish_no_logo (Currently 1): If set to 0, Comskip will be less reliant on logo detection. This can increase aggressiveness but also lead to more false positives.

Important Considerations:

  • Testing: After making any change, process a short segment of your recording to test the results. Analyze the output (EDL or other format) to see if Comskip is correctly identifying commercials without cutting into the program content.
  • False Positives: Aggressive settings increase the likelihood of false positives. You may need to fine-tune settings based on your specific recordings and their characteristics.

Recommended Approach:

  1. Start small: Make one change at a time, test, and observe the effects.
  2. Prioritize: Focus first on min_commercial_size and logo_threshold.
  3. Iterative Refinement: Gradually adjust other parameters based on your observations.

Remember to create backups of your original INI file before making changes. The forum posts I linked earlier may offer additional insights and specific INI files optimized for various situations.