LAV Filters - DirectShow Media Splitter and Decoders

[dansrfe]

What are the recommended values for the mix levels? The defaults have 0.71, 0.71, and 0 for LFE but if I have a 2.1 system that cleanly separates out the LFE frequencies then what should I set that value to? What should I set the other values to as well and are there any particular standard values?

EDIT: I'm getting plenty of bass response from the subwoofer at LFE mix level of 0... I'm confused. Is that mix level slider to mix in more than a particular value of LFE?

[SamuriHL]

Well....I may agree on a multi purpose PC but 100% I'd roll back to Win7 if I could not get it to work on a dedicated HTPC (pretty big limitation IMO and few other advantages for Win8)

I'm not rolling back. It was a HUGE PITA to upgrade to W8. I'll figure it out...even if I have to do custom profiles.

[fairchild]

For 2.0 mixing (to headphones) I'm just using the default settings of 0.71 (center), 0.71 (surround), 0 (LFE) and haven't had any problems, though another user had said the correct/accurate settings are 0.71 (center), 1.0 (surround), 0.71 (LFE)

[e-t172]

Oh, it's that time again when everyone is throwing around their own (often wrong) downmixing coefficients. Reminds me of the famous LFE debate I had with nevcairiel.

To put an end to the never-ending discussions on this topic: the authoritative reference is ITU-R BS.775 - Multichannel stereophonic sound system with and without accompanying picture (http://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.775-3-201208-I!!PDF-E.pdf).

We can see on page 11 that proper downmixing from 5.1 to 2.0 is:

L = 1*L + 0.7071*C + 0.7071*LS
R = 1*R + 0.7071*C + 0.7071*RS

Note that some audio formats allow for metadata containing custom downmixing coefficients. I don't think LAV Audio uses them though, which is unfortunate.

LFE is the most controversial part. See Annex 7 (page 13) on the ITU document for a wealth of information on the subject. There is even a whole section for "education" on how to reproduce LFE correctly, which is surprising to find in a specification. The bottom line about LFE is:

- If you don't have a system capable of really good bass response (i.e. expensive subwoofer), then don't keep LFE. Otherwise it is likely to overload your woofers and result in increased distortion. That's not a good deal in most cases.

- If you do have a system capable of handling high-SPL bass output, then do keep LFE in the mix. LFE is supposed to be played back 10dB louder than any of the other channels. In the bass region signals coming from two stereo speakers are likely to be coherent, which means the output from the two speakers sum up to +6dB (and not +3dB as I believed until recently - although I'm not sure about headphones), and besides, if you're using bass management to a subwoofer (2.1 system), then it is definitely +6dB because bass signals will be summed electrically in the bass management system. So that means LFE must be amplified by 4dB (10 - 6) before being mixed into the two stereo channels. The result is:

L = 1*L + 0.7071*C + 0.7071*LS + 1.5848*LFE
R = 1*R + 0.7071*C + 0.7071*RS + 1.5848*LFE

To be on the safe side the result would require normalization to prevent clipping.

[Nevilne]

I would love to know what's the LAV equivalent for ffdshow default 5.1>2.0 downmix.

[bugmen0t]

@Nevilne
It is:
L = 1*L + 0.7071*C + 1*LS + 0.7071*LFE
R = 1*R + 0.7071*C + 1*RS + 0.7071*LFE
i.e. Center Mix Level = 0.71, Surround Mix Level = 1 and LFE Mix Level = 0.71.

@e-t172
Thanks for these infos. Can you please explain how the 1.5848 is computed? I understand that 0.7071 ~= 1/sqrt(2) comes from solving 2*LFE^2 = 1.

EDIT: Thanks e-t172, i see. It is 1.5848 ~= 10^(1/5). But -3dB is 10^(-3/20) |= 1/sqrt(2). Why is that?

EDIT2: It seems they use 1/sqrt(2) because routing to n channels needs a 1/sqrt(n) coefficient to achieve same loudness (only about -3dB by coincidence; got that from ac3filter site).
Adding 10dB means multiplying by sqrt(10) (from e-t172s link below). Correcting for two channels should be sqrt(10)*1/sqrt(2) = sqrt(5) then, however they seem to use four channels giving sqrt(10)*1/sqrt(4) = sqrt(10)/2 ~= 1.5811. Someone able to explain why?

[e-t172]

@e-t172
Thanks for that infos. Can you please explain how the 1.5848 is computed? I understand that 0.7071 ~= 1/sqrt(2) comes from solving 2*LFE^2 = 1.


0.7071 is simply -3dB. 1.5848 is +4dB. Here's a handy calculator (http://www.sengpielaudio.com/calculator-gainloss.htm).

[alexacolor]

DXVA acceleration of H.264 for video cards AMD broken. It is forever?

[Snowknight26]

DXVA acceleration of H.264 for video cards AMD broken. It is forever?

It's not broken..

[CiNcH]

@nev,

the DVBViewer does not detect AAC audio inside an ADTS container when LAV Splitter is being used. As a result, the DVBViewer does not try to use the AAC decoder selected within its options. The reason is that LAV Spitter proposes a proprietary(?) media subtype GUID:

MEDIASUBTYPE_ADTS_FOURCC = '{53544441-0000-0010-8000-00AA00389B71}'; //ADTS LAV Splitter


DVBViewer GE will add this for the next release. But wouldn't it be better if LAV Splitter proposed the more standard-like MS media subtype GUID:

MEDIASUBTYPE_MPEG_ADTS_AAC = '{00001600-0000-0010-8000-00AA00389B71}'; //ADTS Microsoft

?

[nevcairiel]

LAV Splitters outputs several different types for AAC, one being that one, but also an official Microsoft one (MEDIASUBTYPE_AAC). DVBViewer should be looking at all media types offered for a pin, and not only the first (which is really a special format just for LAV Audio, but needs to be the first so its actually used over the others)

[SamuriHL]

Nev, just wanted to let you know I managed to fix the DTS-HD MA decoding issue. Not 100% sure what fixed it. But it's working for me on W8 Pro X64 with MC18 so I'm good.

[Rimsky]

Because nobody is helping nevcairiel with LAV development. How about you? You could learn programming, so you can help out with LAV development.
Thanks, but I can't
I'm stupid in programming

[dansrfe]

Latest build of LAVFilters doesn't work with the latest build of ICL13 build of MPC-HC x86. If I use the lite version if works fine.

[alexacolor]

Originally Posted by alexacolor
DXVA acceleration of H.264 for video cards AMD broken. It is forever?
It's not broken..

HD2400. latest drivers. in settings dialog set "DXVA (native)" and "h.264" is on, but he says ''active decoder - avcodec", and video freezes.
MPC-HC and Splash player plays h.264 video with hardware acceleration.

[nevcairiel]

I don't have access so such ancient AMD hardware, so i cannot test myself, however i have had no reports of this not working before. Note that LAVs DXVA is only designed to work on Windows Vista/7/8, if you're still on XP, it will most likely not work.

[alexacolor]

I don't have access so such ancient AMD hardware, so i cannot test myself, however i have had no reports of this not working before. Note that LAVs DXVA is only designed to work on Windows Vista/7/8, if you're still on XP, it will most likely not work.

it was a laptop, with windows 7 Home. All other videos plays with Acceleration.
Tomorrow i test this on other ati\amd videoadapters.

[6233638]

Oh, it's that time again when everyone is throwing around their own (often wrong) downmixing coefficients. Reminds me of the famous LFE debate I had with nevcairiel.OK, I have deleted my post now, because I'm really not sure what things should be at. My post was based on the information you posted last time this came up, however: http://forum.doom9.org/showpost.php?p=1597794&postcount=12660

We can see on page 11 that proper downmixing from 5.1 to 2.0 is:

L = 1*L + 0.7071*C + 0.7071*LS
R = 1*R + 0.7071*C + 0.7071*RSHere, I had been using 1.0000 for the surround channels, because this post (http://forum.doom9.org/showthread.php?p=1587652#post1587652) on the subject seemed to indicate that was required for 0.7071 due to how LAV Audio handled downmixing.

However, I have been doing some testing on that, and it seems that using 0.7071 is the correct value to use after all.

- If you do have a system capable of handling high-SPL bass output, then do keep LFE in the mix. LFE is supposed to be played back 10dB louder than any of the other channels. In the bass region signals coming from two stereo speakers are likely to be coherent, which means the output from the two speakers sum up to +6dB (and not +3dB as I believed until recently - although I'm not sure about headphones), and besides, if you're using bass management to a subwoofer (2.1 system), then it is definitely +6dB because bass signals will be summed electrically in the bass management system. So that means LFE must be amplified by 4dB (10 - 6) before being mixed into the two stereo channels. The result is:

L = 1*L + 0.7071*C + 0.7071*LS + 1.5848*LFE
R = 1*R + 0.7071*C + 0.7071*RS + 1.5848*LFE

To be on the safe side the result would require normalization to prevent clipping.This part is confusing to me, because I'm not quite sure I understand how you are arriving at it.

It's accepted that LFE is supposed to be played back at +10dB
If you are sending LFE to both stereo channels, you need to reduce that by 3dB, correct? (same reason Center is at -3dB)

So does that mean each channel is supposed to be amplified by +7dB, or +6dB?

7dB + 7dB = 10dB
Or is the target 9dB (10dB - 3dB = 9dB) which means two channels of 6dB each. (6dB + 6dB = 9dB)

It seems like the channels should be boosted by 7dB, which would be 2.24 LFE, not 1.5848.

0.7071 is simply -3dB. 1.5848 is +4dB. Here's a handy calculator (http://www.sengpielaudio.com/calculator-gainloss.htm).Are you sure this calculator is correct? It says -3dB is 0.707946 and not 0.7071


Anyway, rather than keeping this all theoretical, I decided to create test tones and dump the audio once it was passed through LAV's mixer.

So to do this test, I created a 5.1 FLAC file in Audacity that had L, R, C, Ls, Rs each playing a 500Hz Square Wave tone at 0dB for 2s each, with LFE encoded at -10dB: http://www.sendspace.com/file/4y0wqh (Plays L, R, C, LFE, Ls, Rs)

I then tried different values in LAV's mixer. Setting Center or Surround to 1.000 played back those channels at 0dB, and not -3dB as it had previously been suggested. They need to be set to 0.7071

With the LFE mix, LFE needs to be set to 2.2387 for a 7dB boost (using the calculator you posted) which matches LFE volume with the Center & Surround, at -3dB.
http://www.abload.de/thumb/2.23878ei3t.png (http://www.abload.de/img/2.23878ei3t.png)

1.5848 LFE is noticeably lower in volume. (6-8s)
http://www.abload.de/thumb/1.5848lpeke.png (http://www.abload.de/img/1.5848lpeke.png)


Here is something that I am wondering though: Does LAV's Normalization reduce volume too much?

If I enable Normalization, and use a test file that has all channels at 0dB (http://www.sendspace.com/file/tvmmg5), rather than LFE at -10dB, there's still 6.4dB of headroom left. You can't go above 0dB, can you?

http://www.abload.de/thumb/norm8jljx.png (http://www.abload.de/img/norm8jljx.png)

So it seems like you could go quite a bit louder than that without clipping. But this is not an area where I have any experience at all, so there's a good chance that I am wrong.

EDIT: Nevermind - of course when you start adding channels together in the downmix, they get louder, so this is required to prevent clipping when all channels are at 0dB together.

[GrandeBoma]

Nev, just wanted to let you know I managed to fix the DTS-HD MA decoding issue. Not 100% sure what fixed it. But it's working for me on W8 Pro X64 with MC18 so I'm good.

Perhaps you could tell which folder you have installed it in, and which version of the dll it is. Please?

[e-t172]

This part is confusing to me, because I'm not quite sure I understand how you are arriving at it.

It's accepted that LFE is supposed to be played back at +10dB
If you are sending LFE to both stereo channels, you need to reduce that by 3dB, correct? (same reason Center & Surround are at -3dB)

Not quite. That's what you may think, but there's a little more to it. My previous post that you linked to is wrong, because some of my assumptions were incorrect at the time (hey, we're all learning here).

The reason why center is 3dB is because playing back a signal from two incoherent acoustical sources result in +3dB SPL. Stereo speakers are incoherent sources above bass frequencies (I'm not sure why - I think it has to do with subtle phase differences and room reflections). The most important frequencies (i.e. voice) are not bass frequencies, so stereo speakers will not reproduce them coherently. This is the reason why the ITU is using 3dB for the center channel (I think).

I have no idea why they're using 3dB for surround instead of the more intuitive 0dB. Probably because they don't want surround signals to overwhelm the front stage. In this way the 3dB number for surrounds seems much more arbitrary.

For LFE however, this is a whole other story. LFE is bass only. This means that if you use both stereo speakers to reproduce these frequencies, they will be coherent, and so they will add up to +6dB SPL (and not +3dB - that's the mistake I made in the posts I wrote some time ago). In addition, if in the end you're using a subwoofer to reproduce these frequencies (2.1 system with bass management), then +6dB is definitely the correct gain because the lowpassed L and R signals will be mixed electrically before being sent to the subwoofer, and electrical sum is +6dB.

If you don't believe me, you can easily check this using a measurement microphone and a swept tone signal, comparing L, R, and then L + R together. You will see that L+R start at +6dB in the bass frequencies, and then decreases steadily to +3dB as the frequency increases.

So, here's how you get to +4dB for LFE:
- Add +10dB for LFE so that it is 10 dB louder than any of the other channels.
- At this point, if you send this signal untouched to both L and R, you will end up with +16dB (10dB + coherent acoustical sum from the two speakers)
- So you need to compensate by attenuating by 6dB.
- Result: 10 dB - 6 dB = 4dB, or 1.5849.


Are you sure this calculator is correct? It says -3dB is 0.707946 and not 0.7071

Good question. I think the actual answer is 1/sqrt(2) (0.7071), which is -3.01dB. A difference of 0.01dB is inaudible, which makes this a moot point. Good catch, though.

[SamuriHL]

Perhaps you could tell which folder you have installed it in, and which version of the dll it is. Please?

It's in the lav folder and it's 1.1.0.8.

[bugmen0t]

This means that if you use both stereo speakers to reproduce these frequencies, they will be coherent, and so they will add up to +6dB SPL (and not +3dB - that's the mistake I made in the posts I wrote some time ago).
Is this equivalent to assume there are four channels instead of the actual two as i did in my edited post above?

[6233638]

Not quite. That's what you may think, but there's a little more to it. My previous post that you linked to is wrong, because some of my assumptions were incorrect at the time (hey, we're all learning here).

The reason why center is 3dB is because playing back a signal from two incoherent acoustical sources result in +3dB SPL. Stereo speakers are incoherent sources above bass frequencies (I'm not sure why - I think it has to do with subtle phase differences and room reflections). The most important frequencies (i.e. voice) are not bass frequencies, so stereo speakers will not reproduce them coherently. This is the reason why the ITU is using 3dB for the center channel (I think).

I have no idea why they're using 3dB for surround instead of the more intuitive 0dB. Probably because they don't want surround signals to overwhelm the front stage. In this way the 3dB number for surrounds seems much more arbitrary.

For LFE however, this is a whole other story. LFE is bass only. This means that if you use both stereo speakers to reproduce these frequencies, they will be coherent, and so they will add up to +6dB SPL (and not +3dB - that's the mistake I made in the posts I wrote some time ago). In addition, if in the end you're using a subwoofer to reproduce these frequencies (2.1 system with bass management), then +6dB is definitely the correct gain because the lowpassed L and R signals will be mixed electrically before being sent to the subwoofer, and electrical sum is +6dB.

If you don't believe me, you can easily check this using a measurement microphone and a swept tone signal, comparing L, R, and then L + R together. You will see that L+R start at +6dB in the bass frequencies, and then decreases steadily to +3dB as the frequency increases.

So, here's how you get to +4dB for LFE:
- Add +10dB for LFE so that it is 10 dB louder than any of the other channels.
- At this point, if you send this signal untouched to both L and R, you will end up with +16dB (10dB + coherent acoustical sum from the two speakers)
- So you need to compensate by attenuating by 6dB.
- Result: 10 dB - 6 dB = 4dB, or 1.5849.Well I don't know whether this is correct (I'd have to read up on speaker coherency) but I trust your judgement on that.

However, this now brings up a further question for me: how does this apply to headphones?

I don't use speakers at all these days, and it seems like headphones would either always be coherent, in which case the center channel needs to be -6dB rather than -3dB, or never coherent, which means that the LFE channel needs to be -3dB rather than -6dB.

Good question. I think the actual answer is 1/sqrt(2) (0.7071), which is -3.01dB. A difference of 0.01dB is inaudible, which makes this a moot point. Good catch, though.If I had to guess, I'd say that's probably done for some sort of clipping protection?

[SamuriHL]

To be clear, when you say it's in the LAV Folder, you have it in C:\Program Files (x86)\LAV Filters\x86 and not C:\Program Files (x86)\LAV Filters?


No, on my machines it's in:

c:\users\Samuri\appdata\roaming\J River\Media Center 18\Plugins\lav

:)

To be CLEAR it goes in the same dir as the LAVAudio.ax filter file.

[e-t172]

Is this equivalent to assume there are four channels instead of the actual two as i did in my edited post above?

I'm not sure I understand the question. The +10dB number for LFE has nothing to do with downmixing at all; in fact, it is applied even on full 5.1 systems. It has to do with dynamic range relative to main channels. The ITU recommendation explains this in great detail.

However, this now brings up a further question for me: how does this apply to headphones?

I don't use speakers at all these days, and it seems like headphones would either always be coherent, in which case the center channel needs to be -6dB rather than -3dB, or never coherent, which means that the LFE channel needs to be -3dB rather than -6dB.

I have no idea how headphones behave in this regard. I don't think, however, that the concept of coherence is relevant for headphones. The real question is perceived loudness: if you take a X dB signal and send it to both headphone transducers (as opposed to just one side), how is it perceived in terms of loudness? X? X+3dB? X+6dB? X+10dB? I have no idea. There's probably studies on this though, we just have to find them.

If I had to guess, I'd say that's probably done for some sort of clipping protection?

You mean for 3dB on surrounds? Perhaps it's one of the reasons, yes.

[GrandeBoma]

it turned out i was using 1.1.0.0 version apparenty too old for win8

after switching to 1.1.0.1 I obtained what I wanted

[SamuriHL]

That'd do it.

[bugmen0t]

My question was about EDIT2 of this post (http://forum.doom9.org/showthread.php?p=1611925#post1611925).
I'm not sure I understand the question. The +10dB number for LFE has nothing to do with downmixing at all;
Yes, therefore i didn't asked about that. Applying the +10dB to the LFE is just multiplying it by sqrt(10) (states a table of your link (http://www.sengpielaudio.com/calculator-gainloss.htm)).

However when routing one channel to n others you need to compensate for loudness change by applying a 1/sqrt(n) factor (from ac3filter site). That's where the 1/sqrt(2) ~= 0.7071 comes from when downmixing the center level from e.g. 5.1 to stereo.
So for LFE it should be the sqrt(10) factor for applying +10dB and 1/sqrt(2) because of routing one channel to two, giving sqrt(10)*1/sqrt(2) = sqrt(5). However what they seem to use is sqrt(10)*1/sqrt(4) = sqrt(10)/2 ~= 1.5811 which essentially means compensating for four channels i.e. doubling the actual two.

For LFE however, this is a whole other story. LFE is bass only. This means that if you use both stereo speakers to reproduce these frequencies, they will be coherent, and so they will add up to +6dB SPL (and not +3dB - that's the mistake I made in the posts I wrote some time ago).

So here you say one needs +6dB instead of +3dB, i.e. doubling the latter. That seems to be mathematically equivalent to compensate for four channels instead of two as said above.
I guess this is because of the coherence but i don't understand why it is like that and how to compute that. Routing to one speaker (mono) doesn't need any compensation (states ITU) because of missing coherence, i guess. But e.g. routing LFE to four channels, will it need a sqrt(10)*1/sqrt(2*4) = sqrt(5)/2 factor because the speakers bass will add up pairwise? How does that work?

[MokrySedeS]

I think I found a bug in dxva2cb decoding. This file (http://www8.zippyshare.com/v/45151885/file.html) is stuttering on my gts450.
0.50.1 is the last version that works.

[nevcairiel]

The file plays perfectly for me. Smooth 50 fps (after deinterlacing), even if its only 3 seconds long.

[MokrySedeS]

Sorry, it works now... I had to reset the settings. Though it's still a mystery to me why 0.50.1 worked without a reset :confused:

[e-t172]

Yes, therefore i didn't asked about that. Applying the +10dB to the LFE is just multiplying it by sqrt(10) (states a table of your link (http://www.sengpielaudio.com/calculator-gainloss.htm)).

However when routing one channel to n others you need to compensate for loudness change by applying a 1/sqrt(n) factor (from ac3filter site). That's where the 1/sqrt(2) ~= 0.7071 comes from when downmixing the center level from e.g. 5.1 to stereo.
So for LFE it should be the sqrt(10) factor for applying +10dB and 1/sqrt(2) because of routing one channel to two, giving sqrt(10)*1/sqrt(2) = sqrt(5). However what they seem to use is sqrt(10)*1/sqrt(4) = sqrt(10)/2 ~= 1.5811 which essentially means compensating for four channels i.e. doubling the actual two.

So here you say one needs +6dB instead of +3dB, i.e. doubling the latter. That seems to be mathematically equivalent to compensate for four channels instead of two as said above.
I guess this is because of the coherence but i don't understand why it is like that and how to compute that. Routing to one speaker (mono) doesn't need any compensation (states ITU) because of missing coherence, i guess. But e.g. routing LFE to four channels, will it need a sqrt(10)*1/sqrt(2*4) = sqrt(5)/2 factor because the speakers bass will add up pairwise? How does that work?

Either I'm not understanding you, or you're really making things more complicated that they really are. The fact that sqrt(10) = +10dB is a pure coincidence. I don't know why ac3filter uses the sqrt(2) notation instead of just saying +3dB (which is the same thing) like a sane person. Maybe that's because they're looking at it from the sound *power* perspective (as opposed to the sound *pressure*, which is not the same thing), and they're trying to equalize power instead of pressure, which is a weird way of doing it (and AFAIK invalid, because power summing is only valid for incoherent signals). So let's just forget about sqrt() notations and just use decibels to avoid confusion.

Basically, I'm postulating that in the bass range (and only in the bass range), the same signal coming from two equidistant acoustic sources in a typical room will be coherent at the listening position, meaning that they will add up to +6dB, not +3dB. You seem skeptical, and it is true that right now I only have my microphone measurements to back that up. I should probably search for some authoritative reference to put an end to the speculation.

One thing I'm 100% sure, however, is that +6dB *is* the right number if you're using a 2.1 system with bass management: indeed, in such a system bass from the two channels are summed electrically (as opposed to acoustically), and electric (voltage) sum of two identical signals definitely *is* +6dB.

[turbojet]

Considering LAV fixed a crash when moving to extended display. Could the problem of MPC-BE >1596 auto pausing when a video is opened on an extended display when LAV and madvr 85.x is used?

Changing one of those things makes it play like it should. Reporting it to madvr and mpc-be has led to dead ends.

[JMGNYC]

I'm having a odd problem. I've got a DVD rip into an .mkv via MakeMKV. It's MPEG video.

Using the latest LAV with MPC-HC and madVR it does not play using None as the Hardware Decoder but does play using Intel QuickSync (HD4000 graphics). Anything I can look at to see why this would be the case?

[sneaker_ger]

You could upload a sample for starters.

[supercoolman]

I have a problem when using LAV Splitter playing one MKV file as shown below. LAV Splitter loads both audio and play both of them at the same time... how do I select just the audio track from mka and disable audio track in the mkv file?

a.mkv (video+mp3)
a.mka (flac)

[nevcairiel]

LAV Splitter has no such feature, it is your player loading both files.

[supercoolman]

unfortunately, MPC-HC doesn't see any audio track, but I got two instance of LAV Splitter. one loads the mkv file, the other loads just the mka file. would be good to be able to disable audio in one of the LAV Splitter instance.

[namaiki]

Do you have the Audio Switcher enabled in MPC-HC? (View-> Options-> Audio Switcher)

[supercoolman]

it's disabled.

I tried another mkv file with one video stream and two audio track. MPC-HC can only see one video stream and the audio track I selected with LAV Splitter.

[CiNcH]

LAV Splitters outputs several different types for AAC, one being that one, but also an official Microsoft one (MEDIASUBTYPE_AAC). DVBViewer should be looking at all media types offered for a pin, and not only the first (which is really a special format just for LAV Audio, but needs to be the first so its actually used over the others)

You also propagate the proprietary media subtype GUID...

MEDIASUBTYPE_ADTS_FOURCC = '{53544441-0000-0010-8000-00AA00389B71}'; //ADTS LAV Splitter
... via ppmt, when looking at the available audio tracks via 'IAMStreamSelect.Info'. This method does not provide the possibility to cycle through the media subtypes, right? So the application has to know this specific media subtype GUID in this case.

[nevcairiel]

TBH, IAMStreamSelect.Info is hardly the correct function to try to probe for a media subtype, especially because it doesn't let you provide more then one, which is a very common use-case to support different filters that expect different things.

In any case, i cannot change this without sacrificing some functionality in LAV Audio, so i will not.

[mastrboy]

Will the LAV Splitter support MKV Segment linking/Ordered chapters sometime in 2013?
It's the only feature I miss before I can uninstall the Haali splitter.

Or is this a feature that you nevcairiel will never work on, and it's up to someone else to write the code for this to be implemented?

[MokrySedeS]

@masterboy
http://forum.doom9.org/showpost.php?p=1609160&postcount=13700
:)

[mastrboy]

@masterboy
http://forum.doom9.org/showpost.php?p=1609160&postcount=13700
:)

Awesome, i tried searching this post after "segment" and "segment linking", but didn't try "ordered chapter".

Thanks... no need to reply to my other post then.

[filler56789]

Hmmm, only today :o I noticed that the option "Enable Video Stream Parsing" does not exist anymore. :confused:
Could someone please explain why?

[nevcairiel]

Hmmm, only today :o I noticed that the option "Enable Video Stream Parsing" does not exist anymore. :confused:
Could someone please explain why?

Because turning it off should never be required, or its a bug.

[dansrfe]

LAV Filters (x86) does not work with the 1CL3 builds (or at least the latest 1CL3 builds) of MPC-HC x86.

[nevcairiel]

Then don't use ICL builds. :)

If it works everywhere except in one specific build of one player, maybe you should look there and not here?

[wanezhiling]

Whats the benefit of ICL builds? Any faster?