LAV Filters - DirectShow Media Splitter and Decoders

[nevcairiel]

Haha. So you ran into exactly the same problem... :p Try the Haali Media Splitter, the problem occurs much less there.

Here's how to quickly solve it:
(1) don't flush in NewSegment, flush in EndFlush instead :p
(2) don't send any frames downstream, until you get the first decoded "AVFrame->key_frame" with a positive timestamp

I don't do anything yet to avoid the problem, there is alot of things i plan to do. ;)

PS: I only get one NewSegment. :)

LAVVideo.ax(tid 1004) 4924 : ::Receive: size: 33357, time: 26276276
LAVSplitter.ax(tid f6c) 5031 : ::SetPositions() - seek request; current: 274560000; start: 1148731497; stop: 0; flags: 0l
LAVSplitter.ax(tid f6c) 5031 : ::DeliverBeginFlush on Video Pin
LAVVideo.ax(tid f6c) 4935 : BeginFlush
LAVSplitter.ax(tid f6c) 5037 : ::DeliverBeginFlush on Audio Pin
LAVSplitter.ax(tid f6c) 5077 : ::DeliverEndFlush on Video Pin
LAVVideo.ax(tid f6c) 4977 : EndFlush
LAVSplitter.ax(tid f6c) 5078 : ::DeliverEndFlush on Audio Pin
LAVSplitter.ax(tid 840) 5078 : ::DeliverNewSegment on Video Pin (rtStart: 1148731497; rtStop: 2376126130)
LAVVideo.ax(tid 840) 4978 : NewSegment - 1148731497 / 0
LAVSplitter.ax(tid 840) 5078 : ::DeliverNewSegment on Audio Pin (rtStart: 1148731497; rtStop: 2376126130)
LAVVideo.ax(tid 1004) 4988 : ::Receive: size: 42744, time: -2168267


Have you tried all the various dynamic reconnection methods listed here?

http://msdn.microsoft.com/en-us/library/dd388731%28v=vs.85%29.aspx

QueryAccept fails, as does ReceiveConnection.
According to the documentation, ReceiveConnection would be the right thing, because i need a bigger buffer. But all i get is VFW_E_TYPE_NOT_ACCEPTED

[JarrettH]

Now MPC has decoders, LAV Video has decoders, madvr has decoders, and ffdshow :p

[madshi]

I don't do anything yet to avoid the problem
Yeah, I just spent the better part of two days working on that (and some related things). Still not sure I got it 100% nailed. But I've already found one situation where I'm better than CoreAVC now (CoreAVC shows seeking artifacts, I don't, with one sample).

QueryAccept fails, as does ReceiveConnection.
According to the documentation, ReceiveConnection would be the right thing, because i need a bigger buffer. But all i get is VFW_E_TYPE_NOT_ACCEPTED
Yeah, I would have expected ReceiveConnection to work, too.

Now MPC has decoders, LAV Video has decoders, madvr has decoders, and ffdshow :p
Yeah, well, libav makes it sooooo easy... :)

[nevcairiel]

Yeah, I just spent the better part of two days working on that (and some related things). Still not sure I got it 100% nailed. But I've already found one situation where I'm better than CoreAVC now (CoreAVC shows seeking artifacts, I don't, with one sample).

There is always a balance between finding a fast and a perfect solution. Decoding all frames until ffmpeg indicates that its artifact-free is one solution, but not fast, as it requires CPU time to decode all those frames. On the other hand, analyzing the frames before decoding (at least for H264) looking for key frames and recovery points is a fast solution, but not always 100% reliable.

I'll probably test alot of things as well.

[clsid]

Other common video formats are:
FLV1, VP6, SVQ3, H.263, Fraps, Theora, TSCC, Indeo.

Wishlist:
* Ability to configure which formats are decoded.
* Ability to open config through a shortcut.
* Support for all formats for which FFmpeg has a stable/complete decoder. Less common formats could be disabled by default.
* Optional tray icon (same for the other filters). To get quick access to settings. To view input/output details as mouseover tooltip. As a visual confirmation that the filter is being used. Possibly also show graph details in a submenu.

Please also add support to your filters for reading initial settings from HKLM. That is useful for pre-configuring the filter. So settings precedence order becomes: HKCU > HKLM > Internal default.
Please store audio format settings similar to how it is done for the splitter instead of a single binary value.


Do you have any plans for adding some filtering capabilities in the (distant) future once the decoder is fully working? Stuff like deinterlacing, sharpen, resize, postprocessing, possibly avisynth.

Can you consider making a VFW wrapper in the future? Decoding only would be fine. It probably should not be a lot of work. Many (video-editing) programs unfortunately still use VFW.

Having the above two features would help migrate more people from ffdshow to your decoder.

[madshi]

There is always a balance between finding a fast and a perfect solution. Decoding all frames until ffmpeg indicates that its artifact-free is one solution, but not fast, as it requires CPU time to decode all those frames. On the other hand, analyzing the frames before decoding (at least for H264) looking for key frames and recovery points is a fast solution, but not always 100% reliable.

I'll probably test alot of things as well.
I have one sample where there are sequence headers and a recovery point before every I frame. But still, when seeking there are artifacts if I don't decode the preroll frames. Don't understand why, but that's the way it is. CoreAVC shows artifacts, so does Divx, but ffdshow does not, because ffdshow also seems to decode the preroll frames (without passing them on). This all seems to mostly apply to h264, though. I've added code to really only begin decoding at the first key frame for MPEG2 and VC-1. Seems to work fine. Just not for h264. <sigh>

[Xaurus]

Nev, thank you so much!

With that said, I am greatly confused by all this.

What would be the order of everything in mpc-hc?
lavcuvid->lavsplitter->ac3filter->lavvideo->mad-vr ?

[VipZ]

Awesome work nev, decoder works very well already :)

EVO, VC1 decoding is now available to everyone, your decoder works perfectly for this.

In terms of formats supported, clsid mentioned pretty much all of them, but mpeg1 would also be nice if possible and deinterlacing would be useful as well.

WMV, still has AR issues, but didn't think much would change here yet, the MS asf reader outputs the AR as below for 1080p in 16:9,
VIDEOINFOHEADER:
rcSource: (0,0)-(1440,1080)
rcTarget: (0,0)-(1440,1080)
dwBitRate: 5000000
dwBitErrorRate: 0
AvgTimePerFrame: 400000

VIDEOINFOHEADER2:
dwInterlaceFlags: 0x00000000
dwCopyProtectFlags: 0x00000000
dwPictAspectRatioX: 5760
dwPictAspectRatioY: 3240
dwControlFlags: 0x00000000
dwReserved2: 0x00000000

Currently LAV Video also has the same issue as ffdshow where it doesn't read the AR info for h264 in mkv. Works fine with MPC Video decoder, sample http://www.mediafire.com/?t1i1fqcj6qhwb4s

I have been using LAV Video for the last day or so already and had no stability issues at all.

[xv]

Works great with madVR 0.66 (with 0.67 it crahses on 4:4:4 files). Thanks!

Wouldn´t it make sense to combine it with lavcuvid? The cuda decoder could be used if format/graphics adapter are compatible, otherwise libav is used?

[Mercury_22]

@Nev can you add like madshi VC-1 i from intel ? cause although we (I) appreciate very much a better decoder we (I) NEED very much a decoder for the missing common formats = VC-1i in the first place
Also no DXVA ?

[SamuriHL]

Nice work, Nev. I'll give it a try a bit later today as I'm curious as to how well it works. Once it matures a little bit we should ask J River to whitelist it for use in MC16.

[madshi]

@Nev can you add like madshi VC-1 i from intel ? cause although we (I) appreciate very much a better decoder we (I) NEED very much a decoder for the missing common formats = VC-1i in the first place
Is the Intel VC-1 interlaced decoder better than the Microsoft one?

[JarrettH]

I wish you'd tell us more of this ultimate plan :confused:

[Mercury_22]

Is the Intel VC-1 interlaced decoder better than the Microsoft one?
I can't tell the difference !
Too bad I can't use madvr usually since I can't get rid of the black clipping when I use it and I've tried every setting in madvr and CCC's video settings :mad:

P.S. Are you aware of the black clipping ? Test file (http://www.multiupload.com/UPPQ3P1MAK)
I can't see the "Blacker than black" with madvr (with evr I need to manually adjust the CCC's video settings (brightness and contrast) to see it,)

[nevcairiel]

Other common video formats are:
FLV1, VP6, SVQ3, H.263, Fraps, Theora, TSCC, Indeo.

I'll see about digging up some samples and adding those.


* Ability to configure which formats are decoded.

Of course, all my other filters have that - just wasn't important for testing.


* Ability to open config through a shortcut.

I want that for all filters, need to figure out how to do it properly.


* Support for all formats for which FFmpeg has a stable/complete decoder. Less common formats could be disabled by default.

It'll eventually get a similar interface as LAV Audio, so that when its used with LAV Splitter, all video can basically be decoded.
For everything else, just name media types, so they can be added as well.


* Optional tray icon (same for the other filters). To get quick access to settings. To view input/output details as mouseover tooltip. As a visual confirmation that the filter is being used. Possibly also show graph details in a submenu.

This is on the TODO, but not sure when i'll get to it.


Please also add support to your filters for reading initial settings from HKLM. That is useful for pre-configuring the filter. So settings precedence order becomes: HKCU > HKLM > Internal default.

Thats certainly a possibility, but not a high priority.


Please store audio format settings similar to how it is done for the splitter instead of a single binary value.

I put it on the list a while ago, just needs to be done.


Do you have any plans for adding some filtering capabilities in the (distant) future once the decoder is fully working? Stuff like deinterlacing, sharpen, resize, postprocessing, possibly avisynth.

Any filters that ffmpeg implements through libavfilter would probably be easy to implement. I'll think about this in the future.


Can you consider making a VFW wrapper in the future? Decoding only would be fine. It probably should not be a lot of work. Many (video-editing) programs unfortunately still use VFW.

I have no plans for this at this time. I know nothing about VfW. Maybe someone else can contribute it?

In terms of formats supported, clsid mentioned pretty much all of them, but mpeg1 would also be nice if possible and deinterlacing would be useful as well.

Right, put mpeg1 on the list as well.


WMV, still has AR issues, but didn't think much would change here yet, the MS asf reader outputs the AR as below for 1080p in 16:9,

Right. ffmpeg seems to be unable to detect the AR in those files. I'll check into it eventually. Not a high priority, though.


Currently LAV Video also has the same issue as ffdshow where it doesn't read the AR info for h264 in mkv. Works fine with MPC Video decoder
Technically, thats not an "issue". It can read the AR just fine, however the AR in MKV files is usually encoded wrong. The container has the right AR, but the stream has it wrong. MPCVideoDec ignores the stream AR, unless you explicitly ask for it. I'll add an option to ignore the Stream AR.

Eventually, i want to add an option to LAV Splitter to simply override the Stream AR with the Container AR, like Haali Splitter does - which would also "solve" this problem from the other end.

Wouldn´t it make sense to combine it with lavcuvid? The cuda decoder could be used if format/graphics adapter are compatible, otherwise libav is used?

One day, it may happen. Its more complex then one might think.

@Nev can you add like madshi VC-1 i from intel ? cause although we (I) appreciate very much a better decoder we (I) NEED very much a decoder for the missing common formats = VC-1i in the first place
I have no plans for this. I do hope for the GSoC project that will hopefully implement VC-1i in ffmpeg. We'll see.


Also no DXVA ?
Maybe one day.

[nevcairiel]

I can't see the "Blacker than black" with madvr (with evr I need to manually adjust the CCC's video settings (brightness and contrast) to see it,)

You're not supposed to see BTB, if your screen is calibrated properly.

[Mercury_22]

You're not supposed to see BTB, if your screen is calibrated properly.

I'm using RGB 4:4:4 Pixel Format PC Standard (FULL RGB) over HDMI
Turn the brightness control up far enough for the 2% strip to be visible and there will be a difference between the blacker than black and video black background. from DVE HD Basics (http://www.videoessentials.com/program_notes.php) (AVFoundry Video Forge Generator pdf)

[madshi]

Too bad I can't use madvr usually since I can't get rid of the black clipping when I use it and I've tried every setting in madvr and CCC's video settings :mad:

P.S. Are you aware of the black clipping ?
madVR itself does not clip, if you set it to video levels. The clipping must be happening elsewhere. The correct setup in your situation should probably be to set madVR to video levels and to tell the ATI drivers to output RGB Full Range. Of course you'll also need to pump up brightness on your display to see BTB because (as nevcairiel says) you're not supposed to see BTB normally.

[Mercury_22]

madVR itself does not clip, if you set it to video levels. The clipping must be happening elsewhere. The correct setup in your situation should probably be to set madVR to video levels and to tell the ATI drivers to output RGB Full Range. Of course you'll also need to pump up brightness on your display to see BTB because (as nevcairiel says) you're not supposed to see BTB normally.
I'm using FULL RGB and with video levels in madvr I can easily see BTB (no need to pump up the brightness) but they are clearly the wrong levels (http://imageshack.us/g/19/evrcp.png/)

[VipZ]

Right. ffmpeg seems to be unable to detect the AR in those files. I'll check into it eventually. Not a high priority, though.


Technically, thats not an "issue". It can read the AR just fine, however the AR in MKV files is usually encoded wrong. The container has the right AR, but the stream has it wrong. MPCVideoDec ignores the stream AR, unless you explicitly ask for it. I'll add an option to ignore the Stream AR.

Eventually, i want to add an option to LAV Splitter to simply override the Stream AR with the Container AR, like Haali Splitter does - which would also "solve" this problem from the other end.


Thanks for the info. Regarding WMV, you mentioned before about the issue with the source/splitter not being able to read the AR, I was mostly interested if this same issue would be whats stopping using LAV Video with the MS ASF Reader which is passing the AR over.

[nevcairiel]

Thanks for the info. Regarding WMV, you mentioned before about the issue with the source/splitter not being able to read the AR, I was mostly interested if this same issue would be whats stopping using LAV Video with the MS ASF Reader which is passing the AR over.

Hm, its supposed to keep the AR from the source filter when it cannot figure it out on its own - i'll run some tests.

[fairchild]

I'm using FULL RGB and with video levels in madvr I can easily see BTB (no need to pump up the brightness) but they are clearly the wrong levels (http://imageshack.us/g/19/evrcp.png/)

As madshi and nev has mentioned, you are not supposed to see BTB when your display device is configured properly.

Using that pattern you are only supposed to see the grayscale pattern in the center and on either side of it, you should see 2 gray vertical bars.

Verified it with 3 different renderers and all display the exactly the same (EVR, Haali, MadVR). If you raise the brightness on the display to a non-calibrated level to see if your display chain has BTB or not.

In the pictures you posted, the EVR has the wrong levels or brightness is set too high same with MadVR and TV levels. The correctly calibrated display is the MadVR with PC levels. Again, I don't know if you artficially increased brightness to check for BTB or not.

[madshi]

I'm using FULL RGB and with video levels in madvr I can easily see BTB (no need to pump up the brightness) but they are clearly the wrong levels (http://imageshack.us/g/19/evrcp.png/)
What do you mean? With your madVR screenshots white is 255 and black is 0 (PC levels), respectively white is 235 and black is 16 (video levels). That are most probably the correct numbers for that test pattern. With EVR white is 252 and black is 2. To me the madVR values look correct and the EVR levels seem to be screwed up.

Edit: fairchild was faster... :)

[Mercury_22]

As fairchild said with EVR CP brightness is pump up to so I ca show you the BTB but with madvr PC levels even if I pump up the brightness to 100% I can't see the BTB
Which is not correct isn't it ?

[jmone]

Being away and unable to test, I have two issues/questions around deinterlacing:
- VC1(i) - till it is added to ffmpeg another filter still needs to be used for all VC1 content as I guess the connection will be made but no video is decoded
- How do you deinterlace with madVR as the renderer instead of EVR?

On the format side, it would be great to add DV-AVI (all that home video footage!)

Thanks
Nathan

[Boltron]

I just got home and am dying to try LAV video! Tried a few mkvs and they are looking good. One issue I ran into though is subtitles. Up to this point I was using ffdshow video and it handled subtitles. What do you guys recommend for a subtitle filter that can handle the standard subtitle types? I tried adding ffdshow subtitle filter but crashes mpc-hc. I use mpc-hc, LAV splitter/audio/video, madVR.

Thanks Nev, I've been hoping for a LAV Video for a while. I might just go send you another donation so you can have a pint on me!

[SamuriHL]

Use MPC-HC's built in subtitle renderer.

[Boltron]

Doh! Looked everywhere except the most obvious place ;)

[zipi]

@nevcairiel,
1. Playing this Aiff file: http://www.sendspace.com/file/2q4apm
in graphedit using Lav Splitter/Source , querying the output pin I am getting:

Major Type: Audio - Sub Type: {AFBC2343-3DCB-4047-9655-E1E62A61B1C5} (LAV internal subtype yes?) - Format: {35189950-CAC9-4C8D-819D-B6FAEE15DD9D}
Is this by design?

2. Does the Wavpack Decoder support the Wavpack Hybrid Mode (.wvc complementary correction file to enable an otherwise lossy .wv to play losslessly, if the file was created in this mode.)

3. I noticed MLP files are not seekable.
Thanks.

[fairchild]

As fairchild said with EVR CP brightness is pump up to so I ca show you the BTB but with madvr PC levels even if I pump up the brightness to 100% I can't see the BTB
Which is not correct isn't it ?

This is exactly what happens with me when I use PC levels with MadVR. It clips BTB for me. So use Video levels in MadVR (16-235) plus Full RGB in CCC and you get full BTB and WTW. In your screenshot of TV levels with MadVR you are getting all black levels, so just calibrate your system to the correct brightness and you are set as far as that goes.

[Midzuki]

3. I noticed MLP files are not seekable.

Patches will be welcome. :)
You can wrap MLP in Matroska though.

Program streams support MLP audio, however I don't know if ffmpeg or Mencoder can multiplex it correctly. :(

[nevcairiel]

- VC1(i) - till it is added to ffmpeg another filter still needs to be used for all VC1 content as I guess the connection will be made but no video is decoded

Right, you cannot detect if content is interlaced from a media type.

- How do you deinterlace with madVR as the renderer instead of EVR?

You use ffdshow as a post-processor. Deinterlacing is one post-processing task that i will probably allow sooner or later in LAV Video. ffmpeg already has a neat yadif filter.


On the format side, it would be great to add DV-AVI (all that home video footage!)

I can add DV decoding, some samples would be neat.



1. Playing this Aiff file: http://www.sendspace.com/file/2q4apm
in graphedit using Lav Splitter/Source , querying the output pin I am getting:

Major Type: Audio - Sub Type: {AFBC2343-3DCB-4047-9655-E1E62A61B1C5} (LAV internal subtype yes?) - Format: {35189950-CAC9-4C8D-819D-B6FAEE15DD9D}
Is this by design?

If there is no other type present, then thats a bug. There should always be one "normal" media type, so decoders other then LAV Audio can also work with it. Just need to figure out what is the right type, though!


2. Does the Wavpack Decoder support the Wavpack Hybrid Mode (.wvc complementary correction file to enable an otherwise lossy .wv to play losslessly, if the file was created in this mode.)

Probably not.



3. I noticed MLP files are not seekable.


Thats right. FFmpeg lacks duration/seekign support for MLP files. I would wrap them in MKV to solve this problem - otherwise you can try to open a bug report at ffmpeg/libav and see if anyone wants to fix it in there.

[madshi]

As fairchild said with EVR CP brightness is pump up to so I ca show you the BTB but with madvr PC levels even if I pump up the brightness to 100% I can't see the BTB
Which is not correct isn't it ?
Actually it *is* correct. If you ask madVR to output PC levels, by definition black is set to 0 and white to 255. BTB would be negative numbers with PC levels, and that's not technically possible. So with PC levels the darkest color your display can get is exactly black. This is how PC levels work. If EVR still shows BTB even with PC levels, this is incorrect behaviour by EVR.

TVs and projectors usually expect TV levels. With TV levels, black is set to 16 and white is set to 235. So BTB values are transported to the display in the 0-15 range. Your display will not show the BTB range by default, but you can make it visible by increasing brightness.

madVR is behaving perfectly accurate. It's EVR that is doing things wrong. You need to decide whether you want to use PC levels or TV levels. If you use PC levels, BTB is automatically clipped with no way to get it back by increasing brightness on your TV. If you want to be able to see BTB by increasing display brightness, then you need to use TV levels. Of course you will need to calibrate your TV accordingly, so that BTB and WTW areas are not shown.

[nevcairiel]

Just a small FYI for everyone, i'm moving from GitHub to GoogleCode, now that Google offers native Git hosting.

I already created issues for alot of my TODO points on Google Code, if you think i'm missing anything crucial, feel free to open a new issue as well.
I'll move the repository later today, and then close the GitHub project eventually.

Project: http://code.google.com/p/lavfilters/
Issues: http://code.google.com/p/lavfilters/issues/list

[pankov]

Looking good, nev
:)
One issue that I didn't see but didn't want to spoil your list with my not so good English was about the ordered chapters and "editions" in MKV.
Did you miss it on purpose or it's already implemented and it's actually me that have missed it?

[nevcairiel]

One issue that I didn't see but didn't want to spoil your list with my not so good English was about the ordered chapters and "editions" in MKV.

Added now.

[Xaurus]

Nev,

So from what I understand, LAV Video is an _alternative_ to LAV CUVID.

So what to choose?

I have an Intel 980x @ 4 GHz and an Nvidia GTX 570.

Multi-threads for LAV Video sound sweet, and with some nice
processing options like in ffdshow (sharpering, deband etc) it
would be all we need, with Madvr.

I wish it would be combined though, into one single LAV release.
:thanks:

[jmone]

Nevcairiel - a few things:

1) 4 x 10sec Sample DV Files (thanks for looking at these) : http://www.megaupload.com/?d=FSU2SS9Y
- 16:9 (anamorphic 4:3) in Type 1 and Type 2
- 4:3 in Type 1 and Type 2

2) Deinterlacing: The addition of YADIF would be a big requirement IMO to make LAVVideo a reasonable alternative to FFDSHOW for interlaced material.

3) VC(i): In GSOC we trust (but I'm not holding my breath)!

4) Addition of Decoding with LAVCUVID: I too think that with LAVVideo it makes sense to use LAVCUVID where available as it seems to be a higher quality option for Decoding of H264, VC-1, MPEG2 and MPEG4-ASP including Full Adaptive Deinterlacing and Frame Doubling.

Summary: At present in MC16 we have the ROHQ configuration that uses LAVCUVID if it is possible else it defaults back to FFDSHOW+YADIF for all other types in conjunction with madVR. The results are good and playback is stable. To successfully replace FFDSHOW in this environment, LAVVideo would need to have the same broad decoder support + High Quality Deinterlacing.

Thanks
Nathan

[nevcairiel]

If anyone is good at graphics (drawing and stuff), i could use a logo and a set of icons (4 icons, for LAV Splitter, LAV Audio, LAV Video and LAV CUVID -- i'm fine with the same icon in different colors, or something easy like that). This is your chance to contribute, even when you cannot code. :)

I really have nothing specific in mind, just something that somehow says "LAV", maybe with a subtext of "Filters", somewhat simple and modern. I'll leave everything else to your imagination.
Icons should be at least 32x32, but still look good in 16x16 for tray icons. For the logo i prefer something that is not too huge, and preferably wider then high.

If you have any ideas, and want to help by contributing some artwork, please do contact me. =)

[rsd78]

Hi Nevcairiel,

Any possible solution for subtitle control (similar to Media Control) when using the native WMC player? I don't use MPC-HC and would love to use your software but I would really like to have a way to control/pick subtitle/audio streams.

Thanks

[BloodySword]

If anyone is good at graphics (drawing and stuff), i could use a logo and a set of icons (4 icons, for LAV Splitter, LAV Audio, LAV Video and LAV CUVID -- i'm fine with the same icon in different colors, or something easy like that). This is your chance to contribute, even when you cannot code. :)

I really have nothing specific in mind, just something that somehow says "LAV", maybe with a subtext of "Filters", somewhat simple and modern. I'll leave everything else to your imagination.
Icons should be at least 32x32, but still look good in 16x16 for tray icons. For the logo i prefer something that is not too huge, and preferably wider then high.

If you have any ideas, and want to help by contributing some artwork, please do contact me. =)

Hi, I could make some with Inkscape, I think.
I will make some notes for ideas together with my kind girlfriend and we will make some SVG-Files that you can convert to .ico. But I can make .ico files, as well. =)

Small question: Ho can I force DirectShow to use LAV Source filter instead of FileSourceAsynch to open files wich are open for writing? I remember there were some registry keys where I have to paste the CLSID of the source filter into the key "Source filter". Where can I find them? Thank you.

[VipZ]

Nev, could you add Fraps support to LAV Video

Thanks

[nevcairiel]

Nev, could you add Fraps support to LAV Video

http://code.google.com/p/lavfilters/issues/detail?id=12

[Sven75]

DVCP (Apple DVC-Pro) support would be great as well.

I was about to upload a collection of not yet working clips to test for you, but I just checked and saw that you added a whole lot of new media types. As soon as you provide us with a new release I will recheck my files and upload only those which are still not supported/not working.

When you lose a bit of your momentum in the development of the new decoder, I will come back to haunt you with my PowerPoint 2010 issues as they are still here when using your splitter and decoder. ;)

In any case, I am so excited to see that finally a real alternative/replacement to ffdShow is shaping up. Fantastic work!

[nevcairiel]

DVCP is already working.

[Sven75]

Perfect!

Could you please provide a new release?

It would probably take me ages to setup my VS2010 and all the rest to compile it myself. Should definitely get more into c++ dev instead of just c#...

[VipZ]

http://code.google.com/p/lavfilters/issues/detail?id=12

Thanks :)

Perfect!

Could you please provide a new release?



Here you go, http://www.mediafire.com/?j1jrt3wj6ajp7kr

[Sven75]

Oh, many thanks! :-)

[roytam1]

Just tested, rm splitting and decoding are taken by LAVF and my rmvb clips become choppy (same as ffdshow)

[nevcairiel]

Just tested, rm splitting and decoding are taken by LAVF and my rmvb clips become choppy (same as ffdshow)

http://code.google.com/p/lavfilters/issues/detail?id=26

@everyone, FYI
This will be the new way of things, i'll manage every task/bug/whatever through the tracker. Before, i lost the overview so quickly, it wasn't really working for me anymore.
Now, you can directly subscribe to the issue ("star" it), and get notified when any changes are made regarding your problem/request. I guess most people will already have a google account.

Of course you're free to post issues yourself to the tracker as well, as long as you're sure what you want, and can provide the important facts. :)