Lav Filters

[babgvant 143]

Which file do I modify. I don't remember which one I tried, but when I removed AAC from it, nothing changed. Can you point me in the right direction?

 

directshowplayer.xml in the config folder

 

if you remove a codec, I'm not sure what will happen. It might try to render it using DS intelligent connect, or it might just fail.

[Guest asrequested]

I think that was what I tried. It made no difference. AAC just comes out PCM. I'll take another run at it, tonight.

[babgvant 143]

I think that was what I tried. It made no difference. AAC just comes out PCM. I'll take another run at it, tonight.

 

AAC is [lossy] PCM. It is just waiting in a more convenient way.

[Notfast 7]

Ffdshow can process AAC to AC3 on the fly, very well.

Yes, but ffmpeg is ready installed on the server.

[Guest asrequested]

Babgvant, how do I alter the file to let me use an external filter?

Edited August 15, 2016 by Doofus

[Guest asrequested]

Notfast, I only want this in ET. I have an HTPC with 7.1 Atmos/dts-X. But I guess if someone was streaming to another device in the network and it was connected to a similar system, they might also want what I'm asking for.

[babgvant 143]

Babgvant, how do I alter the file to let me use an external filter?

 

ATM, I don't think you can. Isolating ET's DS process from the PC was a design goal.

 

IIRC, you wanted a feature to add an audio processor b/w the decoder and renderer. That is still on my todo list, but unfortunately my todo list doesn't just include ET tasks .

[Guest asrequested]

I appreciate that, Babgvant. I know you are very busy. I know Luke's goal was to have ET be self-contained, which I actually like. But I think an audio filter needs to be added. LAV is completely redundant in this. It can only decode, and nothing more. Which is not good for hometheater enthusiasts like myself.

[babgvant 143]

I appreciate that, Babgvant. I know you are very busy. I know Luke's goal was to have ET be self-contained, which I actually like. But I think an audio filter needs to be added. LAV is completely redundant in this. It can only decode, and nothing more. Which is not good for hometheater enthusiasts like myself.

 

Decoding is what the decoder should do. Processing should happen in a different filter. Ffdshow doesn't follow the intended design paradigm because of how it evolved, not because it's the right way to do things.

 

It's also worth pointing out that Ffdshow is kind of a mess because of the way it evolved, both from a stability and supportability perspective.

[Notfast 7]

Notfast, I only want this in ET. I have an HTPC with 7.1 Atmos/dts-X. But I guess if someone was streaming to another device in the network and it was connected to a similar system, they might also want what I'm asking for.

 

I know, but Emby use profiles to determine if transcode is needed, so a new profile for Emby theater that gets the server to transcode AAC, and you also got a solution :-)

Edited August 16, 2016 by Notfast

[Guest asrequested]

Babgvant, I agree with you. I only used ffdshow because it was the only way I could process AAC on the fly, correctly. I'd love it if ET had a better way to do it, and I didn't have to use jiggery pokery to achieve my goal. Sadly, AAC is so prevalent, now, that I have to find a way to deal with it. It's unfortunate, because all I need is for it to be stereo AC3 and my receiver will take care of the rest. If I only had a few of these, I wouldn't even mention it, but I have hundreds. I'm slowly going through them and converting, but it's quite laborious.

[Notfast 7]

@, you could do a batch conversion using ffmpeg, it is possible to probe all your files using ffprobe, and if it has AAC then recode it to AC3.

 

I know it is not a good solution, it would be better if our favourite media player worked like we want it to, but it might get you enjoying your movies again.

Edited August 15, 2016 by Notfast

[Guest asrequested]

That's not a terrible suggestion. I've never really messed with ffmpeg, so it would be fun to learn something new. I'll look into it. Any pointers?

[puithove 208]

I didn't say every setting LAV has... Just that what ET has is a superset of what was available via MBT.

 

What GPU do you have that SW DI produces better results than HW? It's not hard to add addition settings to the backend (xml). If you make a [short] list, I can commit to throwing them in.

 

I'm running a GTX970, but I don't think it's a HW vs. SW quality issue so much.  I don't believe madVR uses the card's built-in deinterlacer but rather his own algorithm.  The real issue with madVR though is that I see it constantly switching deinterlacing on/off during some content streams (much of Live TV from my cable company) which leaves some interlaced artifacts.

 

Yadif handles it better as it's inspecting each frame and making a decision whether or not to apply deinterlacing.  It appears to me that when enabling LAV's implementation of Yadif at the decoder stage, then the rest works so much better.

 

FWIW - I've had the same issue with HW decoding in Kodi (on different boxes with different HW) and have always had to run yadif here as well to get a good result.

 

If you could add that as a configurable option in the XML, I for one would appreciate it.

 

So my "List" would be (and it is short  )

• LAV Video decoder / Software Deinterlacing Algorithm (include all options from the list)

Edited August 16, 2016 by puithove

[babgvant 143]

madVR DI is HW based. Sounds like your TV stream may have a metadata issue (not uncommon).

[puithove 208]

madVR DI is HW based. Sounds like your TV stream may have a metadata issue (not uncommon).

 

Ok.

 

Yes, I worked with a couple Kodi devs a year or more ago and we found the streams that the cable company provides have mixed interlaced and progressive which throws everything out of whack.  Yadif (and w3fdif) handle it just fine because they're doing deeper inspection.

[babgvant 143]

Might be worth having a mess w/ madVR's DI settings. Specifically "if in doubt, activate DI".

 

I'll add the LAV DI setting ASAP.

[puithove 208]

Might be worth having a mess w/ madVR's DI settings. Specifically "if in doubt, activate DI".

 

I'll add the LAV DI setting ASAP.

 

I've tried the possible permutations, no good.

 

Thanks, that'd be fantastic.

[babgvant 143]

Added both the SW DI setting and an audio processor option.

 

If using Ffdshow, you will need to configure the audio decoder to process PCM and whatever you want. For some reason that I don't care to investigate the audio processor (which is really the decoder but w/ a different GUID & name and some options removed) won't load properly.

 

It may take a while for it to work it's way into a beta, that's not my department .

[Guest asrequested]

That's awesome! I will test it out when I get home, or when Luke implements this. Thank you, Babgvant. I really do appreciate your hard work, and putting up with my bellyaching

Edited August 17, 2016 by Doofus

[Notfast 7]

That's not a terrible suggestion. I've never really messed with ffmpeg, so it would be fun to learn something new. I'll look into it. Any pointers?

Some code to get you going http://stackoverflow.com/questions/9985345/how-can-i-parse-ffprobe-output-and-run-ffmpeg-depending-on-the-result

[Guest asrequested]

Some code to get you going http://stackoverflow.com/questions/9985345/how-can-i-parse-ffprobe-output-and-run-ffmpeg-depending-on-the-result

Thank you, I'll check it out, but I may not need it if Luke implements Babgvant's modifications.

[Notfast 7]

Thank you, I'll check it out, but I may not need it if Luke implements Babgvant's modifications.

Let's hope he does.

[puithove 208]

Added both the SW DI setting and an audio processor option.

 

If using Ffdshow, you will need to configure the audio decoder to process PCM and whatever you want. For some reason that I don't care to investigate the audio processor (which is really the decoder but w/ a different GUID & name and some options removed) won't load properly.

 

It may take a while for it to work it's way into a beta, that's not my department .

 

Awesome.  Will keep an eye out for updates.  When it does happen, I'm assuming that on existing installs we'd have to add a setting to the XML - is that right, or will it be inserted automatically so we can just change it?  If we need to add, what is the setting name?

 

Thanks!

[babgvant 143]

Awesome.  Will keep an eye out for updates.  When it does happen, I'm assuming that on existing installs we'd have to add a setting to the XML - is that right, or will it be inserted automatically so we can just change it?  If we need to add, what is the setting name?

 

Thanks!

 

The settings will be added automatically after they are saved through the UI (navigate to the audio/video settings, then back out). Close ET, open the XML to edit.

 

Audio processor is either the GUID or name of the filter to load.

 

SW DI defaults to 0, replace with the number of the scheme you want.

 

 None = 0

        YADIF = 1

        W3FDIF_Simple = 2

        W3FDIF_Complex = 3