[Finished] Improvements to Live TV playback rate matching - for testing (2 Viewers)

[Owlsroost]

Hi,

Here are the logs from the client and server. Yesterday there was a Picture Freeze by almost 30 seconds, after that it went on as if I had pressed pause. But that was the only statue in the evening. I hope you'll find something.

Thanks for the logs - quite interesting.

Next time you have one of the long pauses, can you also attach the TV server logs please - I need to see what is happening at both ends to understand the situation. In your 30s pause situation, TsReader appears to be waiting for a timeshift buffer file data read from the server to complete, but I can't tell from the log what the cause might be.

Do you have anything else (other than TV server) running on the server PC, e.g. anything that might generate a lot of network or disk activity ?

The good news is the TsReader logs don't have any continuity or sync errors, so the data integrity looks OK now

I can see a few issues inside TsReader that are showing up in the log occasionally (due to the different way the timeshift buffer files are behaving with the latest TsWriter), so I'm having a think about that at the moment....

The funny thing is also that sooner were the pixelation on the bottom of the screen, now they are at the top. Does that mean something?

Nothing I can think of. As far as I can tell from the logs, any picture problems are probably due to video data not arriving quickly enough occasionally (not because of continuity errors).

What happens when a video decoder runs low on data etc. is going to depend on how it's designed to handle error situations, so it's impossible to give a general answer.

 

[Palm_Maniac]

The Serverlogs are inside the Zip-File, or not? Today I have a few more tweaks and settings on the server to Windows 10 made which may have been responsible for the Picture Freeze. I'll be watching it.

 

[Owlsroost]

The Serverlogs are inside the Zip-File, or not?

Oops - sorry, they are - I hadn't scrolled down far enough....I'll look at them now...

 

[Owlsroost]

As far as I can see from the logs, the 30s pause happened on the 13th (it's in TsReader.bak), I can't find a long pause in TsReader.log for the 14th.

There is no TsWriter .log or .bak for the 13th - only for the 14th and 15th. So I don't have the server log I need...

 

[Owlsroost]

@Palm_Maniac - can you try the attached TsReader please (v4.1.0.9) ?

I've re-worked some of the timeshift buffer file code, so it should work with the latest TsWriter a little better.

Can you run live TV for a few hours and then attach the logs please ? - I'd like to see how it behaves in your system.

 

[Palm_Maniac]

Only from the client?

 

[Owlsroost]

Only from the client?

Yes, that is OK.

I've run this version for a few hours (connected to TV server via WiFi and powerline networking) and it's worked very well.

 

[Palm_Maniac]

Hello there,

Here are the logs from last night. Next I noticed: When you zap the stream starts absolutely clean. No stutter more in the beginning. Unfortunately, the pixelation is still there, it seems to me but before as if there are fewer. However, this is all based on subjective and only yesterday evening. Let's forget the 30sec Still, because it seems to have been caused by the system.

Why do not we change the computer because it runs perfectly with you? What is different here? Realtek LAN chip, Intel chipset, Intel processor ... but everything is standard. I hope continues.

 

[Owlsroost]

Thanks for the logs - the TsReader log looks fine, except that occasionally there are 'Demux : Audio to render too late' messages, which I suspect might be triggering the video pixelation problems.

I'll create a TsReader.ax with some of the buffer file caching reverted back to how it was originally, so you can see if that is better or worse than v4.1.0.9

 

[Owlsroost]

@Palm_Maniac - can you try the attached TsReader please (v4.1.0.10) ?

I've reverted things so only the timeshift 'index' file (the very small one that does not have .ts on the end) is opened in random-access mode - all the data files should use normal Windows 'adapative' file caching.