LAV Filters - DirectShow Media Splitter and Decoders

[e-t172]

Originally Posted by watchman

Well maybe I'm missing something because I'm not really expert in this stuff, but from I read, I always thought that loudness equalization leads to distortion, which is mostly noticeable in victims of loudness war (probably lot of you heard last album of Metallica - Death Magnetic, which is totally distorted and sounds like a piece of crap).

You can't do dynamic range compression without distortion (almost by definition).

Originally Posted by watchman

Once I tried to use free version of this popular EQ, which had good references on head-fi forums (http://www.aixcoustic.com/index.php?id=13) to boost my bass a little bit (2-3db), and I immediately noticed really high distortion. I don't know if that free version is bugged or what, but after that experience, I stopped experimenting with volume boosts.

An EQ in itself doesn't distort the sound. It can, however, clip the output (depending on the input), which results in distortion. You should try with a lower input volume.

[Andy o]

Originally Posted by watchman

Well maybe I'm missing something because I'm not really expert in this stuff, but from I read, I always thought that loudness equalization leads to distortion, which is mostly noticeable in victims of loudness war (probably lot of you heard last album of Metallica - Death Magnetic, which is totally distorted and sounds like a piece of crap).

Also according to this graph, compression is about making louder parts quieter:

If you're editing, you can do whatever you want with compression, including of course making loud sounds quieter. But on-the-fly compression algorithms don't have to do that. If it's clipped in the original, there's nothing it can do. The algorithm should also be smart enough not to clip itself, so you wouldn't have the problem with Death Magnetic.

Once I tried to use free version of this popular EQ, which had good references on head-fi forums (http://www.aixcoustic.com/index.php?id=13) to boost my bass a little bit (2-3db), and I immediately noticed really high distortion. I don't know if that free version is bugged or what, but after that experience, I stopped experimenting with volume boosts.

That's different cause of two things. First is you're just boosting. The real-time compression algorithms used for watching movies should be smart enough to be able to tell when quiet sounds need boosting. If for instance you have a very quiet sound out of one channel, and a loud one out of another, at the same time, the quiet sound won't get boosted cause it would boost everything, including the loud sounds too, and they could clip.

Second, you're just boosting some frequency range. It had been a known problem with iPods that when using EQ you could get clipping fairly easily. Apple JUST fixed it with the iPod Touch 4 and probably others. Fix was simple, boost frequency, then dial back overall gain.

[hubblec4]

another question.

is it possible to play mkvs with multiple Editons inside?

[NanoBot]

"Editions" and "ordered chapters" are only a different names for the same feature, which is actually not included in the matroska part of the LAV_splitter. Therefore ATM you will have to use Haalis splitter if you need this. Hopefully, this feature will be integrated in the future, but of course it is totally nevcariel choice if and when this may happen.

[kirakami]

any plans for Hardware Acceleration CUDA & DXVA?
only CUVID is included says 'Not Available'

[Xaurus]

Originally Posted by kirakami

any plans for Hardware Acceleration CUDA & DXVA?
only CUVID is included says 'Not Available'

Do you have an Nvidia GPU? If not, it won't work.

[THX-UltraII]

I want to swap my ATI HD5450 for a NVIDIA card for two reasons:

1. Use CUVID and that way:
2. reduce CPU usage. I want to use my CPU for a heavy frame interpolation script with avisynth.

What the best card for CUVID?

[BatKnight]

Originally Posted by THX-UltraII

I want to swap my ATI HD5450 for a NVIDIA card for two reasons:

1. Use CUVID and that way:
2. reduce CPU usage. I want to use my CPU for a heavy frame interpolation script with avisynth.

What the best card for CUVID?

If you can afford it, and want the latest tecnology, I would recommend a GeForce GTX 550 Ti, or higher. (FWIW the GTX550 has 192 CUDA cores)
Here in Portugal, such a card is available for about 125 EUROS.

The GT440 or the GTS450 would also be good alternatives, but are from an older generation.

You may compare other editions here: http://www.geforce.com/Hardware/GPUs

[Nicola]

Hi. I asked this on the CUVID thread but I got pointed here, so here I fire my question.
I'm trying to fix an issue I have with h264 decoding: when video has 9, or 16 ref frames, the seeking will disable the video; I have to seek some more to get the video back on, and it doesn't always work. The resolution of the video doesn't seems to be important, it happens for both dvd or fullhd resolutions.
Using mpc-hc as player, haali as splitter (need it for ordered chapters), ffdshow for audio decoding (exept FLAC), video decoding done via software, for fullhd the cpu (AMD 4cores) usage is less than 20% (mardock scramble), haali or madvr for rendering.
So the question is: what is causing this seek issue? my box horsepower should be more then enough, and there aren't spikes of cpu or gpu usage when it happens.

[roytam1]

Got some feedback regarding my rv34seekkeyframe patch/"hack":
The patch did its job well, but sometimes the audio become muted. When seeking to same point again, the audio recovers.

Actually whats the problem between LAV Audio, LAV Video, and the whole DirectShow graph?

[nevcairiel]

Originally Posted by roytam1

Got some feedback regarding my rv34seekkeyframe patch/"hack":
The patch did its job well, but sometimes the audio become muted. When seeking to same point again, the audio recovers.

Actually whats the problem between LAV Audio, LAV Video, and the whole DirectShow graph?

I doubt that just delaying video frame delivery would cause the audio to stop, because audio actually defines the reference clock, not video.

[e-t172]

…unless you're using ReClock.

[Volfield]

Originally Posted by roytam1

New nightly:
http://roy.orz.hm/lavf-w32-nightlies...ame-c73aa37.7z

diff:
http://roy.orz.hm/lavf-w32-nightlies...e-c73aa37.diff

On this nightly when i set subtitles to advanced and open mp4 file without subtitle my players (pot and mpchc) crash. The same file plays ok when i set default.

[roytam1]

Originally Posted by nevcairiel

I doubt that just delaying video frame delivery would cause the audio to stop, because audio actually defines the reference clock, not video.

Then I wonder if this will happen in MPEG2VIDEO and VC1 as well.

[Andy o]

Originally Posted by Nicola

Hi. I asked this on the CUVID thread but I got pointed here, so here I fire my question.
I'm trying to fix an issue I have with h264 decoding: when video has 9, or 16 ref frames, the seeking will disable the video; I have to seek some more to get the video back on, and it doesn't always work. The resolution of the video doesn't seems to be important, it happens for both dvd or fullhd resolutions.
Using mpc-hc as player, haali as splitter (need it for ordered chapters), ffdshow for audio decoding (exept FLAC), video decoding done via software, for fullhd the cpu (AMD 4cores) usage is less than 20% (mardock scramble), haali or madvr for rendering.
So the question is: what is causing this seek issue? my box horsepower should be more then enough, and there aren't spikes of cpu or gpu usage when it happens.

Have you isolated the issue to LAV Video yet?

[nevcairiel]

Originally Posted by Volfield

On this nightly when i set subtitles to advanced and open mp4 file without subtitle my players (pot and mpchc) crash. The same file plays ok when i set default.

Did you enter any advanced configuration, or was the box empty?

[Volfield]

Originally Posted by nevcairiel

Did you enter any advanced configuration, or was the box empty?

I enter this: jpn:eng jpn: pol pol: off (without spaces after : )

[nevcairiel]

Originally Posted by Volfield

I enter this: jpn:eng jpn: pol pol: off (without spaces after : )

Thanks, should be fixed.

[Volfield]

Originally Posted by nevcairiel

Thanks, should be fixed.

Thanks. Can't wait for new nightly

[SamuriHL]

Oh sure, I JUST got done building. Figures.