[dukester]

What is the status on 3D MVC FramePacking playback on the AmLogic S905?

Still HTAB-only for all I know

So what needs to happen? Do the manufacturers need to implement frame packing support?

Your MVC coding stuff works pretty much perfect, BTW

[wrxtasy]

Can anyone post the output of "adb shell cat /sys/class/amhdmitx/amhdmitx0/disp_cap" from the core?

Code:

root@wetekcore:/ # cat /sys/class/amhdmitx/amhdmitx0/disp_cap
720p
1080i
1080p*
720p50hz
1080i50hz
1080p50hz
1080p24hz
4k2k30hz
4k2k25hz
4k2k24hz
4k2ksmpte

[Koying]

@wrxstasy Thanks. Indeed, the 23.976 and 29.970 modes are not exposed.

[movie78]

What is the status on 3D MVC FramePacking playback on the AmLogic S905?

Still HTAB-only for all I know

OK!

Are you waiting for the Amlogic to release a patch to fix 3D MVC framepacking or something else?

Thanks!

[Koying]

Are you waiting for the Amlogic to release a patch to fix 3D MVC framepacking or something else?

If you really mean "waiting", sure I'm waiting...
If you actually meant "expecting", then no. I haven't heard anything leading me to believe AML is busy with this feature.

Nvidia, otoh, asked us clarifications regarding MVC/3D, so there is tangible hope, there (but it's OT here).

[wrxtasy]

So, it sounds like to have HD audio bitstream w/either CT's or fritsch's builds, you either need to disable amcodec and/or adjust refresh rate? If so, neither are desirable I think as this impacts other things (microstutters every 42 secs, 3D MVC, and just losing amcodec on an AMLogic box, which would seem to be its appeal).

You will not have to go about disabling anything, it will be all Automatic if you select HD Audio Passthrough.
amcodec is not disabled, its the Kernel Frame Rate Automation process that outputs nicely synced 23.976fps video that is disabled so it does not interfere with the HD Audio Passthrough itself.

If above is true, then decode to LPCM w/refresh rate & amcodec both enabled, seems the better choice. And, this was/is already working quite well. All this assuming I'm following correctly .

Yes as CT has said, turn off Passthrough and set then number of Audio channels for Multichannel PCM Audio decoding to be output. You will then get Lossless Audio and nicely synced 23.976fps video output.
Minus your HD Audio Christmas lights on your AVR winking at you

[movie78]

Are you waiting for the Amlogic to release a patch to fix 3D MVC framepacking or something else?

If you really mean "waiting", sure I'm waiting...
If you actually meant "expecting", then no. I haven't heard anything leading me to believe AML is busy with this feature.

Nvidia, otoh, asked us clarifications regarding MVC/3D, so there is tangible hope, there (but it's OT here).

Now that is the best news of the day, the fact that Nvidia is looking into 3D MVC .

Keep us update on your progress.

Thanks!!!!

[fritsch]

I still consider fixing the FrameRate Automation a minor thing (e.g. not too hard to do)*. We had such issues on AMD / Intel in the past years and after they counted these values more properly everything was working fine. Never the less, the current switch to 23.976 hz is a hack and that also needs to be solved.

* Needs someone that can compile the kernel and test a bit.

[ChristianTroy]

In case anyone wants to try to build Core's kernel from https://github.com/christiantroy/android...wetek_core

Code:

mkdir out
cd common
make O=../out/ ARCH="arm" CROSS_COMPILE=arm-linux-gnueabihf- wetekcore_tee_custom_defconfig
make O=../out/ ARCH="arm" CROSS_COMPILE=arm-linux-gnueabihf- -j8 uImage
make O=../out/ ARCH="arm" CROSS_COMPILE=arm-linux-gnueabihf- -j8 modules
make O=../out/ ARCH="arm" CROSS_COMPILE=arm-linux-gnueabihf- wetekcore.dtd
make O=../out/ ARCH="arm" CROSS_COMPILE=arm-linux-gnueabihf- wetekcore.dtb

And the toolchain can be downloaded from here: http://openlinux.amlogic.com:8000/deploy...ihf.tar.gz

[hdmkv]

Now that is the best news of the day, the fact that Nvidia is looking into 3D MVC .

Keep us update on your progress.

Thanks!!!!

+1. All the pestering users did here (including myself) appears to have at least gotten nVidia to take notice & consider it.

[wesk05]

EAC3 fixed: http://mirrors.kodi.tv/test-builds/andro...mirrorlist

Have fun.

Is this a WeTeK Core specific build?

Build ea45ce5 didn't quite work on the Minix U1 or the Shield. On the U1, it played as plain Dolby Digital and on the Shield there was no audio. On both of them, debug log shows FFmpeg decoding error for dependent substream.

http://xbmclogs.com/pklaycq5b - Dolby Digital Plus - Minix U1

http://xbmclogs.com/p18effwe2 - Dolby Digital Plus - Shield

[fritsch]

EAC3 fixed: http://mirrors.kodi.tv/test-builds/andro...mirrorlist

Have fun.

Is this a WeTeK Core specific build?

Build ea45ce5 didn't quite work on the Minix U1 or the Shield. On the U1, it played as plain Dolby Digital and on the Shield there was no audio. On both of them, debug log shows FFmpeg decoding error for dependent substream.

http://xbmclogs.com/pklaycq5b - Dolby Digital Plus - Minix U1

http://xbmclogs.com/p18effwe2 - Dolby Digital Plus - Shield

There is no change for the shield at all - I let the code test yesterday evening in the shield thread. The Shield does not (!) do Dolby Digital Plus at all, it is decoded to LPCM.

Edit: Both logs are from U1.

[wesk05]

There is no change for the shield at all - I let the code test yesterday evening in the shield thread. The Shield does not (!) do Dolby Digital Plus at all, it is decoded to LPCM.

Edit: Both logs are from U1.

Well... it doesn't look like it did that on this build.

The second log shows:
06:16:26 T:1116908160 NOTICE: Running on NVIDIA SHIELD Android TV with Android 5.1.0 API level 22, kernel: Linux unknown CPU family 64-bit version 3.10.61-ge817e80

Time to change your glasses

@danjames92 tested 20160208-e84098e. Are there any differences between 20160208-e84098e and 20160209-ea45ce5?

[fritsch]

There is no change for the shield at all - I let the code test yesterday evening in the shield thread. The Shield does not (!) do Dolby Digital Plus at all, it is decoded to LPCM.

Well... it doesn't look like it did that on this build.

The second log shows:
06:16:26 T:1116908160 NOTICE: Running on NVIDIA SHIELD Android TV with Android 5.1.0 API level 22, kernel: Linux unknown CPU family 64-bit version 3.10.61-ge817e80

Time to change your glasses

@danjames92 tested 20160208-e84098e. Are there any differences between 20160208-e84098e and 20160209-ea45ce5?

No idea - but definitely wrong thread. Post your results with current nightly into the shield thread, please. This will cause maximum confusion else ...

[gaskell1066]

@gaskell
I don't know about that particular skin but performance wise there is an abyss between Core and Play

@fritsch
I never had it working on Core, yesterday when I tried the previous build on S905 (kodi-20160208-76875b1-androidtestv3-armeabi-v7a.apk I guess) I had the DD+ on the AVR but playback was stuttering really badly

Thanks for that but is anyone actually using the Aeon MQ skins on their core and how do they find it