LAV Filters - DirectShow Media Splitter and Decoders

[Plutotype]

I experienced this only if the audio output device in Reclock was the same as the default in Windows and was used in WASAPI mode. After exiting MPC-HC, the Wasapi mode hanged in Windows, so I had to restart all other apps like browser, other players, audio apps to refresh the actual status. Windows somehow can not "tell" to all running apps, that the Wasapi mode ended its session and default audio device should be restored in realtime to shared mode.

Therefore Reclock nicely offers 2 ways od playback. Directsound ( can be the default audio output device - mostly 2.0 or headphones ) and Wasapi. Wasapi can be the HDMI audio ( or any other 24/96 capable surround analog output device - Realtek ALC889 for me ). I switch to Wasapi only if Im going to watch a full lenght movie.

[Andy o]

The Evangelion blu-rays are THD 6.1. I'm heading out to finally get my receiver upgraded, and then I'll have to do some direct connections to my display, I'll have some clips later if you still need them.

[Andy o]

I'm gonna try the Realtek driver, if that doesn't work I'm going to go bother the folks on the ReClock forums about it, or back on AVS.

The Realtek driver has other problems when using WASAPI exclusive in ReClock, so that probably is not a solution even if it fixes your problem, that's why I said "if you're not using ReClock's WASAPI...".

Did you try disabling the MC menu sounds, or setting your Windows mixer to stereo?

[nevcairiel]

The Evangelion blu-rays are THD 6.1. I'm heading out to finally get my receiver upgraded, and then I'll have to do some direct connections to my display, I'll have some clips later if you still need them.

Would be great to get a sample to confirm i'm not breaking 6.1 layout.

[Markolc81]

The Realtek driver has other problems when using WASAPI exclusive in ReClock, so that probably is not a solution even if it fixes your problem, that's why I said "if you're not using ReClock's WASAPI...".

Did you try disabling the MC menu sounds, or setting your Windows mixer to stereo?

My Windows mixer is always set to stereo because if it's 5.1 or 7.1 then stuff Like Netflix and Pandora only come out of the L and R speakers and my AVR would not allow me to add processing (like Pro-Logic II for a Netflix Movie).

But I have set it to 5.1 and 7.1 temporarily and it does not make a difference. :mad:

Also now even if I do have LAV Audio set to bitstream, sometimes Windows audio will be incredibly distorted after I exit playing a movie from MPC-HC...This is getting frustrating :mad:

I'm gonna go tinker some more and let you know what's up.

[jakmal]

BTW, anyone sitting on 6.1 samples that are NOT dts? TrueHD? Other multi-channel formats?
I need to confirm the channel order..

Not dts or "not dts or dts-hd" ?

The sample I uploaded in the bug report here: https://github.com/Nevcairiel/LAVFSplitter/issues/42 : has a 6.1 channel DTS-HD soundtrack.

[nevcairiel]

Not DTS at all, not HD or otherwise, but thanks. .)

I actually have samples for all conceivable DTS channel layouts now, and the only one that does not seem to work properly at all is 7.0. But who builds 7 channel without a LFE track anyway!

[Markolc81]

The Realtek driver has other problems when using WASAPI exclusive in ReClock, so that probably is not a solution even if it fixes your problem, that's why I said "if you're not using ReClock's WASAPI...".

Did you try disabling the MC menu sounds, or setting your Windows mixer to stereo?

Well you pegged it. I finally had a lil more time to play around and disabling MC sounds worked. I kinda miss the sounds in WMC but I'd trade it for a working system that's for sure. Now I can bitstream or have LAV Audio decode and windows sounds work fine.

[SamuriHL]

Oh, right, I forgot about that. I disabled 7MC sounds a long time ago on both my AMD and nVidia machines.

[Andy o]

Would be great to get a sample to confirm i'm not breaking 6.1 layout.

You don't need video, do you? I'll post mka clips.

[CruNcher]

@nevcariel

Multithread Encoding Results (Video only Mainconcept Mpeg-2 Decoder) :(

encoded 4531 frames, 29.04 fps, 25364.98 kb/s (Cyberlink Demuxer 2.0) only max 50% core utilization
encoded 4531 frames, 31.13 fps, 25364.98 kb/s (lav splitter 0.26) heavy core utilization fluctuation
encoded 4531 frames, 42.65 fps, 25364.98 kb/s (MPC-HC Splitter) steady core utilization
encoded 4531 frames, 42.79 fps, 25364.98 kb/s (Mainconcept Demuxer) steady core utilization
encoded 4531 frames, 43.10 fps, 25364.98 kb/s (Arcsoft Mpeg Demux) steady core utilization


I guess this is a result of using still big parts of the single threaded ffmpeg decoder ?

[Andy o]

Here's the TrueHD 6.1 sample (http://www.megaupload.com/?d=WKCB97ZA). BTW, if can anyone tell me if it sounds distorted? I think it's in the original audio (explosions and such).

[madshi]

BTW, anyone sitting on 6.1 samples that are NOT dts? TrueHD? Other multi-channel formats?
I need to confirm the channel order..
I've not seen 6.1 E-AC3. But 6.1 TrueHD should exist. You can create 6.1 WAV and FLAC files by decoding DTS with eac3to, of course.

[nevcairiel]

I guess this is a result of using still big parts of the single threaded ffmpeg decoder ?

I don't understand wtf you're on about. LAV Splitter is a splitter, not a decoder. It can without trouble output more then 40fps, seems like some limitation in your encoder.

I've not seen 6.1 E-AC3. But 6.1 TrueHD should exist. You can create 6.1 WAV and FLAC files by decoding DTS with eac3to, of course.

Thanks, i'll check that. Why didn't this occur to me..
BTW, isn't FLAC only strictly defined to up to 5.1?

[e-t172]

BTW, isn't FLAC only strictly defined to up to 5.1?

Some time ago I got FLAC files encoded from Blu-ray audio in 7.1/96kHz/24bit. Kinda extreme. When I tried to play them, I've found that the only player capable of handling them was AIMP. foobar2000, VLC, mplayer, CoreFLAC and vuplayer all failed to open the file.

[madshi]

isn't FLAC only strictly defined to up to 5.1?
No. FLAC supports 1-8 channels, and custom channel masks via metadata information. I somehow doubt that the ffmpeg decoder supports custom channel masks, though. eac3to and madFlac do, of course.

[nevcairiel]

I know it supports 8 channels, i just read somewhere that standard channel layouts are only defined for up to 5.1, and i didn't know about the layout metadata. Thanks for clearing that up.

[CruNcher]

I don't understand wtf you're on about. LAV Splitter is a splitter, not a decoder. It can without trouble output more then 40fps, seems like some limitation in your encoder.

hmm im not sure yet why but it behaves like that on my SB it loses 10 fps efficiency due to this (cores utilization fluctuates) with roozhous direct264 same happens with cyberlinks demuxer (cores get only 50% utilized never reach 100%). It is reproducible everytime not a 1 time thing always if either lav splitter or cyberlink demuxer 2.0 get used in those encoding chains the cores aren't fully utilized anymore.

http://forum.doom9.org/showthread.php?t=141441

I prepare a bigger test with different decoder and content currently, fully automated to gather results faster, could be some interoperability issue i guess as well between Mainconcepts Mpeg-2 Decoder and Cyberlinks and your Demuxer (maybe content and settings based).

[ontherocks]

Anyhow, i have a Onkyo TX-SR607, and DD+ works just perfectly for me - i would assume its somewhat similar decoding units in those two. ;)

Works fine on my 5450 to an Onkyo 876 also.

Yea I missed that page. The page I saw only listed DD, TrueHD but it didn't have DD+. Awesome. :) No idea why it's not working, either. Works fine with my Pio's.

DD+ good here on LAV->Reclock->HD5670->HDMI->TX-NR905

And my Denon 988. Just got a new AVR though a Yamaha V867 and will see how that pans out.

I have come to think that it may be a EDID issue in my case. If any of you guys could post the real time EDID of your receiver (especially the audio block), that will be great.

...also I've a ton of DD+ stuff from all my HD-DVD discs and these also play fine (though all are 5.1, no 7.1) with LAVAudio doing the decoding.
What you are talking about is decoding via LAV. I am talking about bitstreaming.
Moreover the issue is not that DD+ files don't play. In fact they play flawlessly in my system. The problem is the DD+ light doesn't show up in my receiver. What show up is just DD light (without the + sign, just like regular DD media), which makes me believe that its not bitstreaming DD+ correctly.

[SamuriHL]

I have come to think that it may be a EDID issue in my case. If any of you guys could post the real time EDID of your receiver (especially the audio block), that will be great.


Could be. I suggest visiting the EDID thread on AVSForum and get tulli or Vladd to help you out.


What you are talking about is decoding via LAV. I am talking about bitstreaming.
Moreover the issue is not that DD+ files don't play. In fact they play flawlessly in my system. The problem is the DD+ light doesn't show up in my receiver. What show up is just DD light (without the + sign, just like regular DD media), which makes me believe that its not bitstreaming DD+ correctly.

The easiest way to verify that is to try Andy_O's 7.1 DD+ sample linked to earlier in the thread. If you get 7.1, you're bitstreaming DD+.

[ontherocks]

Could be. I suggest visiting the EDID thread on AVSForum and get tulli or Vladd to help you out.
I know what to do with EDIDs, that's not the problem.
I am also sure that my EDID is correct. I just want to see others EDIDs if anything is unusual.

The easiest way to verify that is to try Andy_O's 7.1 DD+ sample linked to earlier in the thread. If you get 7.1, you're bitstreaming DD+.
My receiver is 5.1, so cannot check for 7.1. The only way to confirm if working correctly I guess is the DD+ light showing up.

[SamuriHL]

Well that's a good time. I don't have things setup on my machine to capture the EDID info right now so I'm no help there. I hope you're able to get this resolved.

[nevcairiel]

E-AC3 is not backwards compatible with AC3. If you have a true E-AC3 stream, and you get audio just fine, then it is decoding E-AC3, no matter what any light on the receiver says. E-AC3 does not have a AC3 "core" or something like that.

The exception are Blu-rays, they have a 5.1 AC3 track, and ship an E-AC3 substream, which contains up to 4 channels, and if supported, replaces the AC3 rear channels with its own, as well as adding up to two additional ones for 7.1. But E-AC3 on Blu-rays is really not used at all, its probably only on the Dolby demo disc, and no-where else.

True E-AC3 content is usually found on HD-DVDs, i dunno where else its really in use.

[ontherocks]

E-AC3 is not backwards compatible with AC3. If you have a true E-AC3 stream, and you get audio just fine, then it is decoding E-AC3, no matter what any light on the receiver says. E-AC3 does not have a AC3 "core" or something like that.
I too think so. May be a bug in the firmware. Since you have an Onkyo TX-SR607 which has the same display as mine, could you tell me what lights show up when you play DD+ files from here http://www.demo-world.eu/trailers/high-definition-trailers.php

its probably only on the Dolby demo disc, and no-where else.Any link which demo disc you are talking about? I would like to have a look.

True E-AC3 content is usually found on HD-DVDs, i dunno where else its really in use.
Do you have any such samples? I would like to play around with them.

[Andy o]

Any link which demo disc you are talking about? I would like to have a look.

The DD+ and TrueHD samples on that page you linked apparently are from that blu-ray, unless a demo HD-DVD came out as well that had those trailers. I don't know if the ones uploaded there have been messed with (remuxed, re-encoded video), but I uploaded the m2ts here (http://www.megaupload.com/?d=A18UUKPN) before.

[noee]

@nev,
Last code looks good for first blush, for DTS 5.1 and 7.1 WASAPI and DS. BTW, does DTS-ES count as a legit 6.1 source?

fwiw, there was a syntax error in dtsdecoder.cpp when I did a pull. Not sure if it was my problem or not, but I haven't messed with the decoder code....

- } else if (header.ChannelLayout == 9 && !header.LFE && header.XChChannelLayout) {
+ if (header.ChannelLayout == 9 && !header.LFE && header.XChChannelLayout) {

[nevcairiel]

fwiw, there was a syntax error in dtsdecoder.cpp when I did a pull. Not sure if it was my problem or not, but I haven't messed with the decoder code....

- } else if (header.ChannelLayout == 9 && !header.LFE && header.XChChannelLayout) {
+ if (header.ChannelLayout == 9 && !header.LFE && header.XChChannelLayout) {

Maybe you hit some weird interim version, i dont have any local changes anymore right now, and everything builds fine.

And yes, of course DTS-ES 6.1 is "legit", i don't see why not.
Although, 6.1 in general is so messed up, every decoder seems to handle it differently. Some put the side channels before the BC, some after, its really a mess.

HDMI seems to want the side channels before the BC channel, so i have to shuffle things around, and cannot trust the decoders. :/

I bet the next version will produce funny problems with some of the more "weird" channel layouts, its just impossible to cover all cases.
Although the default layouts like mono, stereo, 5.1, 6.1 and 7.1 should hopefully still work like before.

Sadly, i cannot test native 6.1 content, as my HTPC and AVR dont seem to be on speaking terms when that format is active.
I'll put a test version up tomorrow when i have the options finished, and hopefully someone else can test the 6.1 content on their setup to see if its going out the right channels. ;)

[jakmal]

Do you have any such samples? I would like to play around with them.

Check the file uploaded here:

https://github.com/Nevcairiel/LAVFSplitter/issues/48

[nevcairiel]

That looks like one of those rare DD+ on Blu-ray files. You wouldn't happen to have gotten that from the Dolby Demo Blu-ray? :p
Looks alot like a demo file, tbh. :)

I haven't actually seen a real BD with E-AC3. Like i said, E-AC3 on Blu-rays is special, and may or may not be supported when bitstreaming. I don't really plan to invest much time into this, as its a practically non-existent format.

[madshi]

FWIW, E-AC3 is used on some Blu-Rays for PiP commentary tracks. That's a whole different thing, though. Those tracks have the same format as HD DVD E-AC3 tracks (no AC3 core).

[nevcairiel]

I also haven't seen those on a live BD, all secondary audio tracks i found are DTS Express.

[jakmal]

That looks like one of those rare DD+ on Blu-ray files. You wouldn't happen to have gotten that from the Dolby Demo Blu-ray? :p
Looks alot like a demo file, tbh. :)

I haven't actually seen a real BD with E-AC3. Like i said, E-AC3 on Blu-rays is special, and may or may not be supported when bitstreaming. I don't really plan to invest much time into this, as its a practically non-existent format.

Yes, it is from the BD demo disc.

The OTT services like Netflix and Vudu are using DD+ for their streams in some cases, so maybe it will start gaining traction for physical media ?

I thought you mentioned that one of the advantages of LAV Audio over ffdshow is the fact that LAV Audio actually bitstreams EAC3 :)

Btw, IIRC, that demo file bitstreamed perfectly with 0.25 and started giving issues only in 0.26 .

[madshi]

I have seen quite a few Blu-Rays with E-AC3 PiP tracks. E.g. Transformers 1, IIRC.

[madshi]

Yes, it is from the BD demo disc.
I wish that stupid BD demo disc would never have seen the light of day. Do you really want nevcairiel to spend hours of his time supporting a format which has only ever been used on exactly 1 demo BD disc world wide? There's not a single movie studio which has *ever* used this format.

IMHO every user who posts a sample from that BD demo disc should be banned for a week. :devil: Just kidding, of course...

The OTT services like Netflix and Vudu are using DD+ for their streams in some cases
I doubt they're going to use the stupid Blu-Ray 7.1 E-AC3 logic, though. They're most probably using the HD DVD 5.1 + 7.1 E-AC3 logic which is supported by LAV Splitter / Audio Decoder just fine.

[CruNcher]

Ahh about the encoding issue the same happens @ fastforward in MPC-HC @ 2x i get only 30 fps max not 50 fps as you would expect from that 25fps .ts (Mainconcept Decoder) so something is locking cyberlink demuxer and lav splitter up to 30 fps seems to only happen with them.

[SamuriHL]

There is NO WAY that broadcasts and other streaming services are going to use BD's implementation. Seriously, I agree here....not useful to spend time on this. If anything, I'd prefer to see the effort spent on getting the Dolby decoder dll's from ArcSoft working if you're going to even bother.

[Andy o]

Is there even a 7.1 E-AC3 in existence using HD-DVD's method?

[SamuriHL]

Not that I've ever seen. And broadcasts are never going to use it, either. They'll stick to 5.1. At least for the forseeable future.

[jakmal]

I wish that stupid BD demo disc would never have seen the light of day. Do you really want nevcairiel to spend hours of his time supporting a format which has only ever been used on exactly 1 demo BD disc world wide?

There is NO WAY that broadcasts and other streaming services are going to use BD's implementation. Seriously, I agree here....not useful to spend time on this. If anything, I'd prefer to see the effort spent on getting the Dolby decoder dll's from ArcSoft working if you're going to even bother.

Sorry guys :) Being a verification engineer IRL, I always try to find corner cases where the design doesn't work.. My main concern with respect to filing that issue was that it used to work previously..

[CruNcher]

No go it always locks to 30 fps no difference what you set, it seems like a hard coded parser force, it happens also with other decoder so it seems to be a lav splitter issue it's interesting that Cyberlinks Demuxer has the same problem, many others don't have this issue so it seems avoidable, it also makes fast forward 2x in MPC-HC impossible :(

[SamuriHL]

Sorry guys :) Being a verification engineer IRL, I always try to find corner cases where the design doesn't work.. My main concern with respect to filing that issue was that it used to work previously..

Understood, but, the real world issue is that Nev's time is limited and if he's screwing around with something that we'll never encounter outside of a demo disc, it takes away from real functionality that we can really use. So yes, it's a nice edge case and as a software engineer myself I appreciate those things, it's just not something that's worth giving time to. I'm sure Nev will feel the same way.

[whurlston]

I've not seen 6.1 E-AC3. But 6.1 TrueHD should exist. You can create 6.1 WAV and FLAC files by decoding DTS with eac3to, of course.

If I remember correctly, the DVE HD DVD has 6.1 DD+ and TrueHD samples. I would need to find my copy though.

[Mark_A_W]

There is no code whatsoever that forces any decoder. Blame ZoomPlayer, its "smart play" is usually not so smart.



nevcairiel

Pankov has done some testing for me, and he figured out that ZP is using the Filter Profile for E-AC3 instead of the Filter Profile for FLAC (for FLAC audio). I had LAV Audio chosen in the E-AC3 Profile - which is why it appeared to be forced to me.

It sounds like a Zoom Player issue. But it only happens with the LAV Splitter - it's a bit murky.

Here's what Pankov said to me:

"I've just tried the file and I do think that there is something wrong in ZoomPlayer.
Even though the media subtype is one that's specified in the FLAC profile it insists on loading the filter configured in the E-AC3 profile (added in v8.00 RC2 on my request) which has a totally different media subtype ({AFBC2343-3DCB-4047-9655-E1E62A61B1C5}). If you press "I" while playing the file you'll see the Media Information window and there at the bottom ZP states that it's E-AC3 !?!?"


So either is a ZP issue, and I apologise, or something about LAV Splitter is sending the wrong media type to ZP or something like that...(I've also posted this in the testing section on the inmatrix forum).

Anyway I'm glad Pankov was able to duplicate my issue, therefore I'm not insane (well, maybe a little...) :cool:


Sorry to give you and madshi the run-around.

Mark

[nevcairiel]

it also makes fast forward 2x in MPC-HC impossible :(

LAV Splitter does not support adjusting the playback rate. 1x, or get out! :p

My main concern with respect to filing that issue was that it used to work previously..

It most likely broke because i added the ffmpeg parsers to re-assemble frames, it may not know how to do that with the dependent E-AC3 substream. Otherwise there are no changes that would affect EAC3, it does not use the new timing logic.

Note that E-AC3 will not work with the MPC-HC splitter, as it claims that its AC-3.

I can run some quick tests on that.

[tetsuo55]

Understood, but, the real world issue is that Nev's time is limited and if he's screwing around with something that we'll never encounter outside of a demo disc, it takes away from real functionality that we can really use. So yes, it's a nice edge case and as a software engineer myself I appreciate those things, it's just not something that's worth giving time to. I'm sure Nev will feel the same way.FYI this was exactly ATI's opinion when they decided not to support any streams beyond the strictest L4.1 specs for DXVA h264, look where that brought us.

Other than that, i agree.

[nevcairiel]

Except, there really are no Blu-rays with their main audio in E-AC3 - with the one exception of the Dolby Demo Blu-ray, where they had to show-off their features - and i don't think there are free tools available to create such streams, unlike H264, which has x264 with which you can encode to any level. :p

Its also not like this is a strict limitation in some hardware chip or something, if there ever surface those discs, for whatever reason, i can still try to fix it.

look where that brought us.

To NVIDIA. :)

[madshi]

FYI this was exactly ATI's opinion when they decided not to support any streams beyond the strictest L4.1 specs for DXVA h264, look where that brought us.
The difference is that high spec h264 files were already widely used when ATI made that decision. Furthermore using higher spec h264 encoding can actually be useful sometimes. Both facts don't apply to Blu-Ray style E-AC3 tracks. They're junk, not useful, have never been used in real life yet and most likely never will be.

[tetsuo55]

I think thats a good benchmark.

Give us a commercial or homebrew sample and work on it can be started.
Imho EAC3 is a bad idea anyway.

[nevcairiel]

EAC3 itself is fine, it basically just extends AC3 from 5.1 to 7.1 and allows higher bitrates, no magic involved. Its similar enough to AC3 for decoders to adapt quickly, better then inventing a new format. Its basically the Dolby variant to the DTS-ES extensions or certain DTS-HD HRA profiles.

What they did to Blu-ray style E-AC3 tracks is another matter. To keep the tracks backwards compatible to AC3 decoders, they invented this stupid way of only using this 4 channel E-AC3 to replace the AC3 side channels, and add back channels.

[pankov]

nevcairiel

Pankov has done some testing for me, and he figured out that ZP is using the Filter Profile for E-AC3 instead of the Filter Profile for FLAC (for FLAC audio). I had LAV Audio chosen in the E-AC3 Profile - which is why it appeared to be forced to me.

It sounds like a Zoom Player issue. But it only happens with the LAV Splitter - it's a bit murky.

Here's what Pankov said to me:

"I've just tried the file and I do think that there is something wrong in ZoomPlayer.
Even though the media subtype is one that's specified in the FLAC profile it insists on loading the filter configured in the E-AC3 profile (added in v8.00 RC2 on my request) which has a totally different media subtype ({AFBC2343-3DCB-4047-9655-E1E62A61B1C5}). If you press "I" while playing the file you'll see the Media Information window and there at the bottom ZP states that it's E-AC3 !?!?"


So either is a ZP issue, and I apologise, or something about LAV Splitter is sending the wrong media type to ZP or something like that...(I've also posted this in the testing section on the inmatrix forum).

Anyway I'm glad Pankov was able to duplicate my issue, therefore I'm not insane (well, maybe a little...) :cool:


Sorry to give you and madshi the run-around.

Mark

Mark, Nev,
the problem turned out to be not in ZoomPlayer itself but in the E-AC3 profile that I've created and spread around.
I've used the incorrect media subtype (the one stated above) instead of the real MEDIASUBTYPE_DOLBY_DDPLUS.
After fixing it everything is working OK.

Nonetheless, it's strange why it happens only with LAV Splitter. At the Inmatrix forum Gleb Egorych noticed that LAV Splitter initially sends MEDIASUBTYPE_FFMPEG_AUDIO ({AFBC2343-3DCB-4047-9655-E1E62A61B1C5}) and then switches to the correct one. Is this intentional?
If there are other affected ZoomPlayer users here is the discussion and the correct E-AC3 profile
http://forum.inmatrix.com/index.php?showtopic=13088