If I am understanding correctly, according to this thread:
MCEBuddy conversion task HDHomeRun H 264 should work successfully.
I know this is an extremely dated thread, but I recently moved MCEBuddy to a new Windows 11 PC and upgraded to the 2.6.5 version. I was using the HDHomeRun Unprocessed Conversion Task on my old MCEBuddy version/PC but wanted to try HDHomeRun H 264. I am still using the old CONNECT tuners. Can you confirm that the H 264 Conversion will remove commercials and playback with subtitles using the HDHomeRun Windows App?
Thanks
Gary
In my testing, I have not been able to get it to work with the HDHomeRun (Windows) App. My media PC is an AMD Ryzen 7, 5700G. The HDHomeRun H 264 Conversion defaults Hardware Acceleration to checked with GPU Any. With that setting nothing would convert, deselected the programs seem to convert but will not playback. I have also tested with the HDHomeRun Unprocessed converting the same programs as the H 264 Task. I was able to playback those programs with subtitles but commercials were not removed. Is there something I need to do with the Comskip options to get the commercials removed? I have not attempted any tweaking of Comskip and am using the default options as installed with v2.6.5.
Yes there’s no issue with commercials and HDHR. You just need to tweak your Comskip depending on your content. The best way to start is to attach your conversion log so we can see what’s going on.
Well I have given up on the H 264 Conversion Task, just too many issues to deal with and I am now back to the HDHomeRun Unprocessed task. It works for the most part although my perspective is that it does not work as good as before these upgrades. I have attached a conversion that first had subtitle issues and I changed to a different comskip ini and this run seems to deal with the subtitle ok, and it certainly removed some of the commercials. There is a long one at about the 27 min mark that it did not remove, even in part. Do you want to see the converted file as well? comskip_Asus28G.ini (5.8 KB)
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.
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:
Start small: Make one change at a time, test, and observe the effects.
Prioritize: Focus first on min_commercial_size and logo_threshold.
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.
Thanks much Goose, as always great guidance. I will incorporate those suggestions into an ini and get back to work. Interesting that your “AI BOT” came up with those suggestions. The Comskip_Claude2 ini was derived from Anthropic’s Claude AI bot. Seems to be a battle of the BOTs . The best cutting performance came from the ini that I had been using for years, the Comskip_Asus28G.ini in my first download post. An example was this past week’s 60 Minutes broadcast, the Claude2 ini did not remove ANY commercials where the Asus28G remove all but the lead in commercials. So for now I will modify my Asus28G ini with suggested min_commercial_size and logo_threshold and see what happens. Thanks again.
Gary
If you can elaborate on what channels or types of recordings (series, movies, news, sports etc) that ini works well on, I can add it to our community repository with some notes.