Skip Problem

[gEd]

Playing the bears here with no major skip issues (ffdshow audio and video) it's an SD file so it shouldn't be too bad.
However near minutes 4,5 and 6 there is visible data corruption on both the PC and PCH and if I try to skip during the pixelization it freezes on the PC.

er.. the 3 samples of "corruption" I see at 4:16, 5:12 6:05 are due to the bears breaking the hidden cameras in the making of the documentary...!
I ran the .ts file through TS Doctor and it found no errors at all.

Code:

2012-02-20 20:09:20.897    [DEBUG][1]    Current position is: 9.375
2012-02-20 20:09:20.897    [DEBUG][1]    Asking for position: 69.375
2012-02-20 20:09:20.899    [DEBUG][1]    Using timing.info, jumping to time 69.375000 (offset 39660104)
2012-02-20 20:09:20.903    [DEBUG][1]    After skipping position reports as: 70.218

2012-02-20 20:09:24.599    [DEBUG][1]    Current position is: 77.375
2012-02-20 20:09:24.599    [DEBUG][1]    Asking for position: 137.375
2012-02-20 20:09:24.599    [DEBUG][1]    Using timing.info, jumping to time 137.375000 (offset 83093368)
2012-02-20 20:09:24.624    [DEBUG][1]    After skipping position reports as: 137.468

[Graham]

er.. the 3 samples of "corruption" I see at 4:16, 5:12 6:05 are due to the bears breaking the hidden cameras in the making of the documentary...!
I ran the .ts file through TS Doctor and it found no errors at all.

I am heartbroken. I thought that I might have produced the file that contained the glitch that would allow Sub to reproduce the problem and find a solution for us all. Bollox.

I did wonder why I hadn't seen any pixellation or wotnot when I watched the recording. I had assumed that the absence of visible artifacts was a symptom of my troubles.

I have two more puters that I haven't yet used in my experiments. Maybe, just maybe, I can find something that can be reproduced (the sound you hear is me not holding my breath).

Thanks to all.

[gEd]

A couple of basic things to iron out:

Do you ever defragment your recording drive ?
What CPU do you have in the system (run NPVR in windows mode, bring up task manager and watch the CPU load when u skip)
download the trial of TS Doctor from http://www.cypheros.de/tsdoctor_beta_e.html
Open your recording
click "save new file" and it will creat a new copy but without some of the unneeded stream from the .ts file. it reduced the file size of your sample by 30% when I did it.
see if that version is any better.

[mikeh49]

Since I have the delayed skip problem on one machine (Lounge machine in my sig), I did some additional testing. My skips are accurate as to time, but delayed for many seconds almost all the time. Either forward or back.

On this system, GBPVR (PVRX2) skips just fine on any file I try - dvr-ms (the best skip response), mpg, or ts recorded by NPVR (acceptable, but not as smooth as dvr-ms). These files are all played from the Videos menu as the box has no tuner; it is accessing recordings over the wired network from the recordings folder of two other boxes with tuners.

NPVR skips OK with the same dvr-ms and mpg files, but handles its own ts files terribly. The first skip may be OK, but today the second froze the screen and only a stop could recover. These are played from the Videos menu.

Regarding CPU use, on GBPVR, it is about 40-50% playing, jumping to 85% on skip in a dvr-ms file. NPVR is 30-40% playing, jumps to 60% on skip of dvr-ms, but goes to 0 when it freezes on a ts file. The system plays smoothly with either GBPVR or NPVR.

Logs from both GBPVR and NPVR attached, testing was 15:30 to 15:50 on 2/20.

One thing of note in the logs is that NPVR inserts an AC3 Renderer filter in the graph, while GBPVR does not.

Maybe this helps?

Should mention that these are all HD MPEG recordings.

[Grampz]

NPVR skips OK with the same dvr-ms and mpg files, but handles its own ts files terribly. The first skip may be OK, but today the second froze the screen and only a stop could recover. These are played from the Videos menu.

I also have this same issue, and it is the last hurdle I am trying to overcome. What baffles me is that there is no consistency. A pause or ff recovery might take 3 seconds, up to 30 seconds, or not at all. I have tried many combinations of audio/video decoders. Only setup i can use to avoid synchronization issues are PVD12, and AC3 Filter. Renderers are EVR and defaults. Any decoders not needed are disabled. Also, using the Haali splitter with it's defaults.

There has to be a timing issue somewhere.

Greg

[sub]

Been busy with work the last few days so I haven't had time to look into it yet, but I'm wanting someone to post an example .ts file so we can both test the same file. That said, I'd prefer not to start those tests with a file that has corruption in it, so I'm hoping for a sample other than the one Graham posted.

[Graham]

Been busy with work the last few days

So busy that you may have missed ...

er.. the 3 samples of "corruption" I see at 4:16, 5:12 6:05 are due to the bears breaking the hidden cameras in the making of the documentary...!
I ran the .ts file through TS Doctor and it found no errors at all.

http://forums.nextpvr.com/showthread.php...post431812

Thanks

[sub]

Sorry, missed that - just grabbed it.

[sub]

Actually...

Can someone reproduce it on a channel that has a ticker that shows the time? This would make testing easier.

[mikeh49]

Here's 50 MB of a clear QAM recording:

http://dl.dropbox.com/u/63603555/Modern%...1-chunk.ts

No ticker though, so I'll see if I can find something.

What size chunk do you need?