LAV Filters - DirectShow Media Splitter and Decoders [Archive] - Page 217

jakmal

4th May 2012, 10:45

The first configuration will need more power because it has to download 60 full frames from the GPU, and then upload them back in madVR. The second configuration only downloads 30 frames from the GPU (and uploads those back in madVR), and the deinterlaced 60 progressive frames never leave the GPU again (its happening all inside madVR)

Awesome, thanks for the explanation! I started experimenting with this because RO-HQ in JRiver MC 17 was dropping too many frames, but I was able to configure things to not drop any frames in MPC-HC with the same madVR configuration (set to deinterlace / HQ madVR settings suggested by Mathias). I turned to Custom mode and set up LAV Video Decoder to Weave, and things were quite smooth after that in MC 17.

Can you advise on whether JRiver should disable deinterlacing fully in LAV Video Decoder in RO-HQ mode? (as long as madVR is set to deinterlace -- Hopefully, JRiver can detect that part of the madVR settings automatically).

6233638

4th May 2012, 12:03

The only thing from that list even active for DVDs would be LAV Audio, which has only gotten very minor internal changes in the last version.
I never understood when DVD with madVR worked and when it breaks, it seems completely arbitrary to me, so i don't consider this a bug, because it probably isn't caused by any of the changes but some other random thing.Yeah, LAV Audio is the only LAV Filter that shows up with DVDs. Interesting though, as ffdshow Audio Decoder also breaks DVD playback with madVR, but LAV has been fine until now. (0.50.1 works with DVDs)

The reason I posted my full filter list, is that I thought it might help you reproduce the problem, but I understand if it's not a priority for you. I just find it far more convenient to have the ability to play back DVDs off the disc, than having to rip them before playing.

nevcairiel

4th May 2012, 12:21

(0.50.1 works with DVDs)
But thats the part that doesn't make sense. The behavior of LAV Audio has not changed at all in 0.50.2, there were just some internal changes that have no impact on things you can measure on the outside really (except when you play stuff with a rate > 1.0)

Not to mention that other people reported that it didn't actually work for them even with LAV.
Its something that needs to be fixed in madVR (if it can be fixed), everything else is just random.

dukey

4th May 2012, 12:40

You can always try the open source dvd navigator. Was one of the main reasons for making it, to get past the macrovision error, when used with other renderers.

jmone

5th May 2012, 08:30

Anyone else seeing VC-1(i) decoding issues on Intel HW (i7-2600K 3000) when using HW Accel? It is fine if I use SW Decoding, but when using Intel HW Accel I see macro blocks in dark areas. This is very noticable in fade in/out transitions where all the black areas all of a sudden break up into large macro blocks. It never happens on light content. I'm currently running V 8.15.10.2656

EDIT - I see there is a newer driver out (2696) so will try that first
EDIT 2 - Looks like driver version V 8.15.10.2696 fixes the issue

egur

5th May 2012, 12:17

Sebastiii

5th May 2012, 14:26

Hi,

I get a crash in ffmpeg.dll (ffdshow) when LAV Video is in graph with NV12 output checked. I get it on my HTPC Win7 32 bits with my Nvidia GTX460 and only when ffdshow video decoder was as Post Process filter with resize function enable.

I cut a little sample that show up the issue on my HTPC :
(http://dl.dropbox.com/u/10536084/lav/Sample/LesI.avi)

It surely related to ffdshow but if i use another video decoder it didn't crash.
If i keep NV12 + rezise OFF -> no crash.
If i uncheck NV12 + resize ON -> no crash
If i enable NV12 + rezise ON -> it crash

I have test on clean setting from LAV and/or ffdshow.
I don't know if it's ffdshow or LAV Video related.

Thanks.

nevcairiel

5th May 2012, 14:29

If ffdshow crashes, its obviously a ffdshow problem.
It always had some issues with NV12 input.

STaRGaZeR

5th May 2012, 14:46

As of 2696 (or 2669 - don't remember). I'm not aware of a single VC1 or WMV9 (wmv3 fourcc) clip that causes artifacts. If someone encounters a bad clip please share. Most corruptions occur in H264 but they are very rare.

Here: http://www.mediafire.com/?4k142w9fofq60dg

Happens since forever.

EDIT: driver 2696.

Sebastiii

5th May 2012, 14:48

I will post on ffdshow thread.
Thanks Nevcairiel :)

Edit : It's really weird, doing the same test on my dev PC with 9600GT (with same driver version) and same LAV/FFDShow build (clean setting) and i didn't get the crash.

nevcairiel

5th May 2012, 15:38

Here: http://www.mediafire.com/?4k142w9fofq60dg

Happens since forever.

EDIT: driver 2696.

I can confirm the macroblocking problem in the fight scene with the Intel decoder.

2696 seems overall pretty solid, 4k decoding finally started working for me and i found a setup to use the dedicated card for output and the iGPU for decoding now, without the weird side-effect of the VGA screen (which always became the primary after a hibernation cycle)
I just use the Intel GPU for audio output to the receiver and my dedicated GPU is directly connected to the TV. As a bonus, i can be sure the receiver is not tampering with the image. :)

egur

5th May 2012, 15:47

Here: http://www.mediafire.com/?4k142w9fofq60dg

Happens since forever.

EDIT: driver 2696.

The block artifacts at the start of the clip is a known issue shared by all QS codecs. A fix is WIP.
The block artifacts in the middle of the clip is a new problem. I'll report this clip.
If you have other clips (H264,VC1,MPEG2,WMV3/9), please share.

SamuriHL

5th May 2012, 16:41

I can confirm the macroblocking problem in the fight scene with the Intel decoder.

2696 seems overall pretty solid, 4k decoding finally started working for me and i found a setup to use the dedicated card for output and the iGPU for decoding now, without the weird side-effect of the VGA screen (which always became the primary after a hibernation cycle)
I just use the Intel GPU for audio output to the receiver and my dedicated GPU is directly connected to the TV. As a bonus, i can be sure the receiver is not tampering with the image. :)

I may go that route myself. Right now I've got it set up so that my 5870 is primary output and Virtu is active so media converter can use QS for transcoding (the only thing I really need QS for). Side note: A file that took an hour to transcode on my 8200 takes 9 minutes with QS on the I5 Sandy Bridge and 4 minutes on the I7 Ivy Bridge. NICE. :)

Sebastiii

5th May 2012, 16:50

JarrettH

6th May 2012, 03:30

But thats the part that doesn't make sense. The behavior of LAV Audio has not changed at all in 0.50.2, there were just some internal changes that have no impact on things you can measure on the outside really (except when you play stuff with a rate > 1.0)

Not to mention that other people reported that it didn't actually work for them even with LAV.
Its something that needs to be fixed in madVR (if it can be fixed), everything else is just random.

I can confirm that 0.50.2 stopped working with DVDs too and I went back to 0.50.1

nevcairiel

6th May 2012, 08:42

I can confirm that 0.50.2 stopped working with DVDs too and I went back to 0.50.1

I just tried, and LAV works just fine with DVDs and madVR for me.

Had to run AnyDVD HD to get rid of the copy protection, but i don't think it ever worked without it.

Trib

6th May 2012, 09:12

Edit: Got it fixed nvm. Ill change this to a thank you post to Nev for continuing to develop this awesome project and all who help him :)

STaRGaZeR

6th May 2012, 10:26

I can confirm the macroblocking problem in the fight scene with the Intel decoder.

2696 seems overall pretty solid, 4k decoding finally started working for me and i found a setup to use the dedicated card for output and the iGPU for decoding now, without the weird side-effect of the VGA screen (which always became the primary after a hibernation cycle)
I just use the Intel GPU for audio output to the receiver and my dedicated GPU is directly connected to the TV. As a bonus, i can be sure the receiver is not tampering with the image. :)

The whole "it needs a screen to be connected for it to work" stuff is horrible. I hope Intel develops something like cuvid, because being tied to DXVA sucks. Good to hear that, because I'll probably face the same issue one day.

The block artifacts at the start of the clip is a known issue shared by all QS codecs. A fix is WIP.
The block artifacts in the middle of the clip is a new problem. I'll report this clip.
If you have other clips (H264,VC1,MPEG2,WMV3/9), please share.

The corruption I wanted to report is the one in the middle, as you say the very first frames are usually corrupted anyway. Good to hear someone is working on it, looking forward to it.

terence_13

6th May 2012, 13:35

nevcairiel, very off-topic, but I need to mention this bleeding obvious fact publically (don't get it wrong):
you need coders to join your project. Its has become to good and important to let it die once you drop the ball (which most coders do after 3-4 years when working as hard as you do). I know you do this just for fun, but your fun has become a stable basis for e.g HTPCs for many people. I guess trying to join forces with people from ffdshow or other projects is not an option?

JEEB

6th May 2012, 14:02

you need coders to join your project. Its has become to good and important to let it die once you drop the ball (which most coders do after 3-4 years when working as hard as you do). I know you do this just for fun, but your fun has become a stable basis for e.g HTPCs for many people. I guess trying to join forces with people from ffdshow or other projects is not an option?
Let's just get this straight:

The project uses Git. Makes it very easy to clone and thus fork, and play with it (as well as make patches that can be passed to the developer in whatever way you wish).
Nevcairiel has not as of yet said 'NO U' to any patches.
Nevcairiel is available here, on IRC and so forth. Getting patches in if one needs such is very easy.
Even if nevcairiel would suddenly one day disappear, the project can be forked very, very easily by people who want or need changes to be done. Once again, "capitalism^Wgit, ho!"
LAV Filters have much less legacy hacks related to the ffmpeg libraries (compare to FFDShow, for example, which bears a rather large amount of legacy from the times of milan, when such hacks were most probably just unavoidable), they're almost completely vanilla. Dropping in and updating things couldn't be much easier.

Do pardon me, but I don't really get what you're throwing your hands around about? If someone wants to join the effort, it is very, very easy to do so. Nevcairiel on his side is doing a damn good job, as well as taken notice on certain things people have possibly done in their own forks of LAV Filters.

tl;dr You do not just have other people "join" your project unless there are specific needs, and/or stuff people want to do (because not like you'll keep them coding with money, either -- not to mention that Git makes general contribution to a project and/or a complete fork in case of SuddenDisappearance much, much more easier). And LAV Filters have already become big enough to be a nice thing to contribute to if anyone wants to.

nevcairiel

6th May 2012, 14:20

Nevcairiel has not as of yet said 'NO U' to any patches.

To be fair, there weren't all that many patches by other people. :D

Anyway, i think my code isn't all that hard to read (unlike ffdshow *cough*), as evidenced by many people "borrowing" chunks of it into their own software (with or without credit), so if someone wants to contribute a new feature or just fixes, they are free to do so.

Midzuki

6th May 2012, 15:45

nevcairiel, very off-topic, but I need to mention this bleeding obvious fact publically (don't get it wrong):
you need coders to join your project.

Hmmm, I think it's Haali who "needs coders to join his project" :sly: :D

clsid

6th May 2012, 16:34

I dream of the day that ffdshow is no longer needed and its functionality is available through better software.
Some functionality for which no (good) alternatives are available yet:
* audio mixer => planned for LAV
* DXVA + subs in basic players such as WMP/WMC that have no custom video renderer => maybe xy-vsfilter will support rendering to an output pin in future
* video processing => needs a new standalone filter
* VFW support (VFW is ancient, but still used by several popular video editors) => needs a simple wrapper for FFmpeg, possibly just decoding, encoding is not important.

6233638

6th May 2012, 16:36

I just tried, and LAV works just fine with DVDs and madVR for me.

Had to run AnyDVD HD to get rid of the copy protection, but i don't think it ever worked without it.Have you done any DLL modifications or installed another DVD navigator on your system?

If not, what's your setup?
It's not working on a stock Windows 7 install for me. (I also run AnyDVD HD)

nevcairiel

6th May 2012, 16:38

Have you done any DLL modifications or installed another DVD navigator on your system?

If not, what's your setup?
It's not working on a stock Windows 7 install for me. (I also run AnyDVD HD)

This Windows is not even a week old, it doesn't get much more "stock" then that. :)

I just grabbed latest MPC-HC and latest ffdshow from http://xhmikosr.1f0.de/, latest madVR, and of course my latest development builds of LAV, and put one of the LotR DVDs in the drive, and voila, it played just fine.

I didn't use ReClock, if that maybe matters for something?

JarrettH

6th May 2012, 18:01

Yeah, LAV Audio is the only LAV Filter that shows up with DVDs. Interesting though, as ffdshow Audio Decoder also breaks DVD playback with madVR, but LAV has been fine until now. (0.50.1 works with DVDs)

The reason I posted my full filter list, is that I thought it might help you reproduce the problem, but I understand if it's not a priority for you. I just find it far more convenient to have the ability to play back DVDs off the disc, than having to rip them before playing.

Same with me, LAV Audio is/should be the only thing working

Anything I can report to help fix it? I get the DVD Macrovision fail even if I run AnyDVD now

Windows 7 32-bit
MPC 4235
ffdshow 4440 with libavcodec decoder

Why do newer installs of LAV want a reboot?

dukey

6th May 2012, 18:11

Is the disc/drive/windows region matched ?

Soukyuu

6th May 2012, 18:59

You can always try the open source dvd navigator. Was one of the main reasons for making it, to get past the macrovision error, when used with other renderers.That's what I did, and it worked exactly ONCE before reverting to the "no sound output even though sound track detected"-behavior I described in my post (http://forum.doom9.org/showthread.php?p=1572537#post1572537) =/

terence_13

6th May 2012, 22:19

...If someone wants to join the effort, it is very, very easy to do so...

That is my whole point: it is easy and still nobody does join.... I fealt inclinded to bring this up, as I am currently trying to revive (meaning: fixing nasty bugs that prevent people from using it) a nice little programm called Media Control that does a splendid job in remote controling FFDshow from Windows Media Center (autorefreshrate, subtitles etc). It used to have a very dedicated coder like nevcairiel, but he ran out of time. I think nevcairiel indicated several times he would appreciate support, but here we are....nothing...

Anyway, did not want to hijack this thread, but I do think LAVFILTERs does deserve at least as many contribiturs as FFDShow (just check over at sourceforge)

clsid

7th May 2012, 00:01

ffdshow only has a few contributors that are still somewhat active. ffdshow has the development pace of an elderly pedestrian, while nevcairiel is cruising on the Autobahn at 200mph.

ryrynz

7th May 2012, 00:06

ffdshow has one major contributor, Haruhiko Yamagata, so that's probably not a good comparison. Clsid and a couple of others fix bugs and perform minor updates from time to time but that project does languish a little.

You beat me to it Clsid. :P

Yes I agree LAV filters has become pretty much THE dominant splitter & decoder package and yes it is surprising that there aren't more developers out there wanting to work on media players/filters and decoders for the community.
However, while there's a desire to tinker and improve, development WILL happen sooner of later, sometimes there just needs to be that tipping point or driving force to get someone motivated enough to start.

TheShadowRunner

7th May 2012, 00:43

Hi Nev,
I'd like to report yet another FLV4/VP6 bug; I think it's a libavcodec issue but not sure.
A sample is here (http://videoff7.free.fr/sample_artefact.flv).
As the video progresses, PQ becomes worse and worse, it can be seen easily on the newscaster's face 10 seconds in.
No such issue happens when decoding with the On2 VP6 directshow decoder (http://videoff7.free.fr/On2_VP6 Decoder.zip).
Any idea?

JEEB

7th May 2012, 01:01

...
Never heard of the app/filter, but sure is still alive (http://mediacontrolmce.svn.sourceforge.net/viewvc/mediacontrolmce/trunk/?view=log), by the original dev too. If it was ever broken for a long amount of time, it meant that people either didn't care about it, or didn't know of it. I don't use the Windows Media Center for one, thus the project would be waaay off my chart of interest (nor do I use remote controls with my PCs). In short: Developers don't come around if they don't have their own interest in it (something in it for themselves).

Also, you're trying to say that no-one else is doing anything for LAV. People have found buggy samples and made up small parts of them that would replicate the problem easily, posted bug reports that should be easy to handle and so forth. Also, a patch that was originally written by Lord for the CCCP got into LAV's mainline in one way or another... is this still "no-one is helping/doing anything at all"? For crying it out loud, people also report bugs in ffmpeg/libav so that things would get fixed in LAV, which uses ffmpeg's code (which also merges changes from the libav side).

Edit: If I weren't busy with my own things I would be checking out libavresample/libswresample APIs to see how to implement a mixer in LAV, although nev had already mentioned he was interested implementing it himself. Unfortunately, I am both busy as well as lack the API knowledge to quickly see how it could be implemented. Derp.

People are working on LAV, and nevcairiel is a very active developer that leaves very little amount of people going for a fix regarding things that have been broken for a while. Why? Because the darn bugs get fixed before that, as you report them. Now, if someone had some project and/or need for LAV that nev wouldn't want to implement himself but would be ready to accept, then you would probably see someone working around in his own fork to make patches that could then be merged into mainline (mainline pushing rights really don't mean much in Git world as work can be easily done without them as well). Put this in other words, should the LAV Filters' development stagnate, and bugs would start to arise, we would most probably see some developer(s) doing work in their own forks.

But I feel that I am talking to a wall...

...
Try with ffplay or by outputting ffvhuff or pngs with ffmpeg (or even with a new'ish mplayer(2) if you wish). Also make sure to turn all sorts of post-processing off from the VP6 decoder to make sure you are getting a non-post-processed picture from the decoder. If the results still don't match up, hit up ffmpeg's trac (http://ffmpeg.org/trac/ffmpeg), see if there is anything similar, and if there is none -- post a new bug report for libavcodec. Otherwise add yourself in a comment for the related bug report.

dukey

7th May 2012, 11:56

yes it is surprising that there aren't more developers out there wanting to work on media players/filters and decoders for the community.

Maybe there just aren't that many talented coders that want to work for free, in their time off :p Even big open source projects are normally done by just a handful of guys, sometimes even 1 guy. Look at media player classic, just made by Gabest basically. It's also not easy to just start working on someone elses project, it can take significant time to understand how other peoples code even works, before you can start to do anything useful with it.

ryrynz

7th May 2012, 13:13

I can say if I had coding experience I'd be helping out, Avisynth script serving.. yes please. Perhaps the software just works well enough to not warrant new people wanting to get stuck in, or perhaps it's the factors you mentioned, there's no doubt many reasons. I will say, it's a little sad when original coders disappear as they never seem to return.. Always wondered why they just up and leave their babies to other people to rear without so much as a helping hand along the way.

XadoX

7th May 2012, 13:16

...while nevcairiel is cruising on the Autobahn at 200mph...
I hope he will not run out of fuel :p
Great work, keep cruising nevcairiel!

ryrynz

7th May 2012, 13:22

Indeed, we are spoilt a little me thinks. One more week and we get Madshi back too.. :P

nevcairiel

7th May 2012, 13:38

One more week, and i'll be gone though.
Can't help it, just have to play Diablo3 all day then. :p

ryrynz

7th May 2012, 13:50

Okay.. nobody tell Madshi about Diablo III.. ya'll be be vewy vewy quiet.

Also I know a guy who knows a guy who's taking a whole month off to play it. :eek:

JustinChase

7th May 2012, 13:59

One more week, and i'll be gone though.
Can't help it, just have to play Diablo3 all day then. :p

Ha! So happy I'm not the only one :) :) I feel a fever coming on that day :scared: I'm afraid I'll have to miss work :sly:

The beta has me all revved up; I can't wait to get past Level 14 and see some new stuff!!

nevcairiel

7th May 2012, 14:08

I'll still go to work, i'll just spend my evenings with it. :p

Trib

7th May 2012, 16:24

I'll still go to work, i'll just spend my evenings with it. :p

Cant say I blame you, gonna do the same :D

JMGNYC

7th May 2012, 17:00

Xaurus

7th May 2012, 18:17

Anyone have a recommendation for minimum CPU/GPU for MPC-HC/madVR/LAV.

I currently am getting about 300 frame drops during a typical BD movie ripped to .mkv with a 23.976fps/bitrate of about 32mbps. I also get a huge number of frame drops playing back a 59fps 720p .mkv. I seem to get fewer frame drops using no HW decoding compared to DXVA Copyback or DXVA Native. I would have expected the reverse.

I currently have a E4200 2.2Ghz/ATI 5570. Is it time for an HTPC upgrade?
I have this one:

http://www.asus.com/Graphics_Cards/NVIDIA_Series/ENGTS450_DC_SLDI1GD3/

No fan. I also use it with a passively cooled Intel Core i3 2120T. Couldn't be more happy, it has played back everything I have thrown at it without problems.

JMGNYC

7th May 2012, 19:54

I have this one:

http://www.asus.com/Graphics_Cards/NVIDIA_Series/ENGTS450_DC_SLDI1GD3/

No fan. I also use it with a passively cooled Intel Core i3 2120T. Couldn't be more happy, it has played back everything I have thrown at it without problems.

Are you using CUVID with LAV with this card or relying on the CPU to decode the video?

THX-UltraII

7th May 2012, 20:21

I have some questions about the configuration settings of my JVC RS45 projector with LAV filters+MPC-HC.

1. HDMI Input
I can select a. Standard (16-235), b. Enhanced (0-255) or c. Super White (16-255) in the JVC RS45 projector menu.
What is the setting that should be used when using a HTPC with:
- Blu-Ray content to play
- Media Player Classic Home Cinema
- madVR as output renderer (set to PC levels 0-255 as RGB output levels)
- LAV video filter (RGB set to 'untouched' in LAV video filter)
- NVIDIA GTX460 with default 1080p23 profile BUT with n vidia fix during installation (for further information about this fix check http://forums.nvidia.com/index.php?showtopic=216099&view=findpost&p=1328716)

2. Color Space
I can select YCbCr (4:4:4), YCbCr (4:2:2) or RGB. It seems that only RGB seems to show the correct colours. I thought Blu-Ray specification was 4:2:2?

nevcairiel

7th May 2012, 20:22

Hi,

I put together a small test version.
I remember someone asking about speex support, and because i put together a cross-compiling environment today, i thought i would just build the speex library and see if it worked.

x86: http://files.1f0.de/lavf/LAVFilters-0.50.2-13-g0e154c3.zip
x64: http://files.1f0.de/lavf/LAVFilters-0.50.2-13-g0e154c3-x64.zip

Other then speex, nothing interesting since 0.50.2

If you can, please try speex decoding. I only had one sample, but that one worked.
Note that most likely only LAV Splitter + LAV Audio will work. If you can provide me with files that contain speex, and an info which other splitter properly outputs it, i can maybe make it work with that as well.

Xaurus

7th May 2012, 20:30

Are you using CUVID with LAV with this card or relying on the CPU to decode the video?
Yes, CUVID exclusively. I've configured my screen for pure 23,976 for all progressive material.

For interlaced 29,970 and 25,000 I use 59,940 and 50,000, respectively. Works like a charm with the hardware deinterlacer provided by CUVID.

TheShadowRunner

7th May 2012, 22:43

Try with ffplay or by outputting ffvhuff or pngs with ffmpeg (or even with a new'ish mplayer(2) if you wish). Also make sure to turn all sorts of post-processing off from the VP6 decoder to make sure you are getting a non-post-processed picture from the decoder. If the results still don't match up, hit up ffmpeg's trac (http://ffmpeg.org/trac/ffmpeg), see if there is anything similar, and if there is none -- post a new bug report for libavcodec. Otherwise add yourself in a comment for the related bug report.

Thank you, I could reproduce the bug with mplayer2 too.
I submitted the bug here (https://ffmpeg.org/trac/ffmpeg/ticket/1282).