Skip Problem

[mvallevand]

If you can post some logs showing this, I'll see if I can work out why.

Sub, I think all of my problems are related to a sample file from bgowland that you may still have linked, related to this thread. http://forums.nextpvr.com/showthread.php...e-channels

2012-03-07 19:16:17.978 [DEBUG][7] first pcr: 0 (at offset 0)
2012-03-07 19:16:17.978 [DEBUG][7] last pcr: 6847820091 (at offset 52428312)
2012-03-07 19:16:17.978 [DEBUG][7] Video Duration: 76086.8899

The problem is the first PTS value (which I think you mean) is deep in the file (not offset zero) and it is 6842858271 which would give the correct 55 second timeline. If nothing else perhaps a sanity check on file size/duration is necessary?

Martin

[sub]

Was this his Real_Hustle-chunk.zip file? Unfortunately I think he's removed it from his server.

This isn't very common scenario, transition from off-air channel, plus no timing.info, so probably not a big deal.

[mvallevand]

Was this his Real_Hustle-chunk.zip file? Unfortunately I think he's removed it from his server.

This isn't very common scenario, transition from off-air channel, plus no timing.info, so probably not a big deal.

True, Timing.Info will probably make this file skippable on the NMT, but it doesn't display any video for me on the PC so it might be worth a look. If bgowland doesn't still have it I can upload it.

Martin

[bgowland]

It should still be there (I think) I'll check.

[bgowland]

Yep still there. I've PM'd a reminder to the link if you do happen to want to take a look sub.

[johnsonx42]

Johnsonx42, can you see if this gets rid of the 20 second delay you're seeing?

Yep, that seems to have done it:

Code:

2012-03-07 19:49:33.330    [INFO][14]    DigitalRecorder.StopStream() handle: 1
2012-03-07 19:49:33.740    [INFO][14]    No more streams active. Stopping device.
2012-03-07 19:49:33.741    [DEBUG][14]    About to request async graph stop
2012-03-07 19:49:33.743    [DEBUG][9]    Graph stopping... (async)
2012-03-07 19:49:33.746    [DEBUG][9]    Graph reports state 'Stopped'
2012-03-07 19:49:33.746    [DEBUG][9]    Graph stopped (async)
2012-03-07 19:49:33.746    [DEBUG][14]    Async stop completed successfully

thanks sub.

[johnsonx42]

Getting rid of the 20 second delay seems to have markedly improved comskip skipping. Recordings made since putting in that patch skip much more accurately; while they're not all exact, many of the skips are about as close to perfect as I could wish for. I think the delay must have affected the recording's timeline or timing.info somehow. Perhaps the timing.info stopped right away, but the recording itself continued for the extra 20 seconds, throwing off the way positions were calculated at playback time?

I think this must explain why I first reported great success with the skip patch, as I was playing a recording that had already been made... then after that when watching new recordings, I reported that skipping was worse than ever.

I do have to qualify this with the observation that the recordings made tonight are not ones that I commonly watch so I can't really say these exact shows skipped badly last week... but I haven't seen any show skip as well as tonight's did for a LONG time. Tomorrow's recordings will tell the tale, but things look good so far... fingers crossed!

edit: well, perhaps that didn't actually change anything. I just went through last night's Star Trek:TNG episode that recorded after I went to bed, and it's comskip jumps were as far off as ever.

[mvallevand]

I have a few files that are like this where TSReader works better than TSReader2 . If you look at the picture the video is just going into a commercial break, but the marked comskip point is quite a bit a ahead. The comskip time @994.93 seem to be correct and when I play it with MPC-HC I see that at 16:43 to so I am assuming the point is valid.

My findings. Without Timing.Info TSReader and TSReader2 seem to behave the same way. With Timing.Info, TSReader2 doesn't work any better but TSReader reports the correct times as I skip and it reacts to the comskip edl perfectly. With the NMT and Timing.Info the skip is in the right spot.

Log attached

Martin

[sub]

I do know of something that could be causing it, which I intentionally decided to leave until the next main release, for fear of breaking binary compatibility.

[Bluethunder]

I don't have the latest and greatest update installed, just skip patch 2 (not 2b). Not sure if this has been caught or fixed (theres alot of pages to this thread now).

I was watching live tv, and missed a bit, so I tried to rewind.

It flipped out on me; hung, and then started playing the same clip over and over again, about 30 second delay.

It stopped responsing to inputs, and I had to ctrl-alt delete to shut it down.

I immediately grabbed the logs, if they are any help.

I am running the colossus, .ts, and AAC sound.