Comskip Skipping Accuracy On Extenders vs Clients

[Taddeusz]

I run Show Analyzer on my server and have noticed a difference in the accuracy of the positions of the skips for HDPVR recordings on my extenders compared to when they are played on my Client or Placeshifter. Skips on either software clients are pretty much dead on. However in comparison, skipping on my extenders appears to be off by a substantial amount.

Is there any way this can be fixed or worked around?

[Brent]

Not sure about placeshifter, but the accuracy seems exactly the same on my HTPC computer versus each of the extenders in my home. I'm not even sure I know why it would be different. I am using comskip though and not ShowAnalyzer

[CollinR]

There is an adjustment in the stv that allows you to specify an offset for skipping. Maybe you have a value stuck in there, this effects all comskip not just from a specific source though.

[Taddeusz]

It's odd to me that there is a difference. But there is. I kind of wonder whether or not it has to do with decoder differences in the extenders vs clients.

The way I noticed this was that I did all my tuning at work via Placeshifter. Then when I got home and started watching stuff on my extenders it was all off by a small amount. Sometimes by enough to really be annoying.

Not sure if adjusting the start and stop offsets would be enough due to some noticed variability.

[stanger89]

Probably semantics, but the difference isn't in comskip, it's in the accuracy of the seeking of the PC vs the extender. I thought at one point Sage added something to help accuracy and there was a property to define how long Sage should try to seek to exactly the right place, but I can't find it at the moment.

-edit

These might be it:
miniplayer/align_iframes_on_seek=true
miniplayer/num_reseek_attempts=3
miniplayer/seek_diff_for_reseek=2000
http://forums.sage.tv/forums/showthr...iff_for_reseek

[bits]

I agree with stranger89 that it is probably not comskip responsible for the difference between server, client or Placeshifter.

FWIW I have semiautomatically cut thousands of commercials using comskip over the last 6+ years and it is not perfect, not by a long shot. However, IMO and in my experience it is more accurate more often over a wider range of TV shows/stations/formats than anything else out there. But the bottom line is that you have to expect the start and end of a commercial segment to be off up to 30% of the time. The larger the variety of TV shows/stations you record the more likely it will be to run into some errors. I recently started using the comskip stvi and I am quite pleased with it. In fact I have pretty much stopped cutting commercials.

What I mean by semi-automatic is that I manually confirmed the commercial segment start and end before batch processing the files and removing the commercials.

[Taddeusz]

Yea, I've never cut commercials and always depended on either the stvi or SageMC to auto skip.

Also, I've always used either Comskip or Show Analyzer. The old Show Analyzer is by far the most accurate. The new one, not so much. Most of that is due to the School House feature available in the old one and the tuning I was able to perform to make it dead-on accurate most of the time. Once the new Show Analyzer has School House I'm sure I can make it just as accurate.

Comskip is ok but it's slower, especially on H.264 video, and more difficult to tune correctly. I've actually stopped using Comskip for H.264 due to it's lack of speed and relative inaccuracy when compared to the new Show Analyzer. I still use the old Show Analyzer for my HDHR recordings because nothing beats it.

[bits]

Originally Posted by Taddeusz

Yea, I've never cut commercials and always depended on either the stvi or SageMC to auto skip.

Also, I've always used either Comskip or Show Analyzer. The old Show Analyzer is by far the most accurate. The new one, not so much. Most of that is due to the School House feature available in the old one and the tuning I was able to perform to make it dead-on accurate most of the time. Once the new Show Analyzer has School House I'm sure I can make it just as accurate.

Comskip is ok but it's slower, especially on H.264 video, and more difficult to tune correctly. I've actually stopped using Comskip for H.264 due to it's lack of speed and relative inaccuracy when compared to the new Show Analyzer. I still use the old Show Analyzer for my HDHR recordings because nothing beats it.

Interesting comment regarding the old SA. I think had you had experience with actual confirmation of cut points on a frame accurate basis you probably would not be of the opinion that it was far more accurate. I did find SA to be quite accurate on certain shows but across a wide variety of shows/show types/channels ect it was not as dependable as comskip nor was it any faster. I did use SA for about a year before switching over to comskip and I have given SA numerous re-tries over the years.

I have not tried comskip on h264 nor SA since I am still in the land of mpeg and will be until I have no other option or when vrd and comskip catch up with it. I will check out the new SA in mean time.

[Taddeusz]

Originally Posted by bits

Interesting comment regarding the old SA. I think had you had experience with actual confirmation of cut points on a frame accurate basis you probably would not be of the opinion that it was far more accurate. I did find SA to be quite accurate on certain shows but across a wide variety of shows/show types/channels ect it was not as dependable as comskip nor was it any faster. I did use SA for about a year before switching over to comskip and I have given SA numerous re-tries over the years.

I have not tried comskip on h264 nor SA since I am still in the land of mpeg and will be until I have no other option or when vrd and comskip catch up with it. I will check out the new SA in mean time.

I have always read that the edl time based cut points, which I use, are more accurate than the frame based cut points. Or am I mistaken?

Edit: Not to mention that the frame based cut points require a preset framerate. Not a problem if all your recordings are the same but if you have mixed 720p and 1080i recordings and set the comskip stvi to 29.97 your 720p recordings won't be right.