Page 24 of 30

Re: Testing MrMC 3.8.0 for Google Play

Posted: 13 Mar 2019, 19:59
by hieyemtim
Thanks for the update and all your hard work! When do we get to taste the next delicious beta?

Re: Testing MrMC 3.8.0 for Google Play

Posted: 13 Mar 2019, 23:30
by Ghostm
koying wrote: 13 Mar 2019, 12:39 Nvidia is pretty vague regarding the root cause, but I read between the lines that it would be a encapsulation (IEC61937) issue with high-bitrate . The fact that they couldn't resolve it, with all their power, means that it is a tricky one, unfortunately, and that it will take a lot of digging into (non-public) documentation to chase it...

PS: from https://club.myce.com/t/atmos-audio-dro ... ead/399840
The actual layout of the Dolby MAT frame used to encapsulate TrueHD for HDMI transmission is unknown and undocumented, so unless someone finds the specification how to do this, I’m afraid there isn’t even a starting point to start looking
So my guess is that the ffmpeg implementation (probably reverse-engineered) is the only one publicly available, hence all software players suffering the same
Thanks for the reply.


If that is the issue with my file there was also discussion here.



https://github.com/google/ExoPlayer/issues/3803

Sounds like lav filters on PC fixed this, but I know this is not a pc we are using as well.

Re: Testing MrMC 3.8.0 for Google Play

Posted: 15 Mar 2019, 03:34
by Ghostm
Just an update on this, i am not so sure this is a max bitrate issue, i posted on the shield forum the lion king which has a high max bitrate plays fine. I am however seeing this issue on seamless branched disney movies only. On any non seamlessly branched movies i do not. That thread koying linked to, i will try dead men tell no tales which was reported to have dropouts, it is however one single mt2s file. Its possible this is fixed already on the shield as far as max bitrate issue goes?

whoever maintains lav filters also recently fixed that issue (seamless branching issue), apparently the windows 10 media player did not experience dropouts for seamless branched movies ripped to mkv, and that was used to fix the bugs with the lav filters truehd decoder from my understanding. The seamless branched disney movies are currently the ones i am seeing dropouts on and where the link koying provided the discussion headed as problematic movies. The file i uploaded to koyings cloud service is a seamless branched movie ripped to mkv. I would love to blame ripping software since oppo players have the same dropouts, but seeing how the movie and apps player built into windows 10 did not (and which apparently lead to the lav fix) now im not so sure.

Re: Testing MrMC 3.8.0 for Google Play

Posted: 16 Mar 2019, 03:54
by Ghostm
OK nevermind, max bitrate issue is definitely real, dead men tell no tales has dropouts on high bitrates.

Really disappointing, I think i might have to go back to my zappiti, no fault of mrmc team, unfortunately this is a shield issue imo unless koying can fix something like this himself on the shield. Unfortunately the shield team still have not sorted this out. Maybe one day. Zappiti has some issues as well but it has workarounds for those.

Hopefully the shield team figures this one out.

Re: Testing MrMC 3.8.0 for Google Play

Posted: 16 Mar 2019, 14:39
by koying
I didn't notice LAV was open-source :)

Apparently, the issue is that we assume a MAT frame is fixed 2560 bytes, while it actually can be more on high bitrate.

For reference, the commit to backport: https://github.com/Nevcairiel/LAVFilter ... 75832468bf

Re: Testing MrMC 3.8.0 for Google Play

Posted: 16 Mar 2019, 14:49
by cosmoxl
koying wrote: 16 Mar 2019, 14:39 I didn't notice LAV was open-source :)

Apparently, the issue is that we assume a MAT frame is fixed 2560 bytes, while it actually can be more on high bitrate.

For reference, the commit to backport: https://github.com/Nevcairiel/LAVFilter ... 75832468bf
Hmmm...is this the same problem that caused MrMC on tvOS to not detect Atmos from some video sources. Davilla said block size was the problem.

Re: Testing MrMC 3.8.0 for Google Play

Posted: 16 Mar 2019, 15:01
by davilla
cosmoxl wrote: 16 Mar 2019, 14:49
koying wrote: 16 Mar 2019, 14:39 I didn't notice LAV was open-source :)

Apparently, the issue is that we assume a MAT frame is fixed 2560 bytes, while it actually can be more on high bitrate.

For reference, the commit to backport: https://github.com/Nevcairiel/LAVFilter ... 75832468bf
Hmmm...is this the same problem that caused MrMC on tvOS to not detect Atmos from some video sources. Davilla said block size was the problem.
no.

Re: Testing MrMC 3.8.1 for Google Play

Posted: 19 Mar 2019, 01:07
by Ghostm
I am not sure how i could get into the beta testing program but it looks like some possible changes to some of the issues i posted. I do currently have a shield and apple tv, although I have 2 more weeks to decide if the shield is going back to amazon.

Re: Testing MrMC 3.8.1 for Google Play

Posted: 19 Mar 2019, 04:54
by hieyemtim
Ghostm wrote: 19 Mar 2019, 01:07 I am not sure how i could get into the beta testing program but it looks like some possible changes to some of the issues i posted. I do currently have a shield and apple tv, although I have 2 more weeks to decide if the shield is going back to amazon.
Hi Ghostm. Just go here: https://play.google.com/apps/testing/tv.mrmc.mrmc

Take a look at the first post in this thread. Once you’ve joined, when a new beta is available it will automatically download on your nvidia shield. I like both the Apple TV4k and the shield but the shield has a lot more to offer us.

Re: Testing MrMC 3.8.1 for Google Play

Posted: 19 Mar 2019, 12:49
by koying
Beta updated to 3.8.1

I also created an alpha version which (hopefully) should fix the TrueHD dropouts.
To join the alpha, you have to join the Google Group at https://groups.google.com/d/forum/mrmc-alpha with the google adress you bought MrMC with, above enrolling as a tester at https://play.google.com/apps/testing/tv.mrmc.mrmc

Not 100% sure that will work, tbh. Just ping if you don't receive the alpha version, which should have a 809b6bf hash.