VC-1 Encoder SDK - Professional released

[Midzuki]

@TinTime

awesome works! :thanks:

will be possibile to add audio support ?

If by "audio support" you mean "encoding to AAC / AC3 / MPA",
then TinTime will have my official approval. :D

[buzzqw]

just wma... and producing a wmv file with both audio and video muxed...

for aac/ac3/mpa.. i prefer eac3to/neroaac... ;)

BHHH

[Midzuki]

Well, I think it wouldn't be terribly difficult to call libaften and lame_enc. :) Anyway, WMA is not suitable for the containers MP4 and M2TS ;)

@ benwaggoner: it seems you missed my whole point.
Even when I use "insane" settings for VC-1 encoding,
duplicated frames will still happen. And when the ASF output is chosen,
the compressor may generate less frames than the framecount of the source file (1095 frames instead of 1100 frames, for example).
I serious doubt the WMV~encoders didn't behave so
before the "VC-1 Era".

[TinTime]

just wma... and producing a wmv file with both audio and video muxed...

for aac/ac3/mpa.. i prefer eac3to/neroaac... ;)

BHHH

I agree - I think eac3to has got all the audio options well covered. Does the audio not work at all now? I'm wondering if I've broken something or whether it never worked properly. I'll give it a go now.


Well, I think it wouldn't be terribly difficult to call libaften and lame_enc. :) Anyway, WMA is not suitable for the containers MP4 and M2TS ;)

I think you're overestimating my C++ skills :)

[TinTime]

I've made a small change to completely ignore audio when the output is .vc1 or .raw. New version is here (http://www.mediafire.com/file/nnmajnyzzng/AVS2ASF.zip).

It now doesn't hang/crash when writing to .vc1 with this script:

Colorbars()
Trim(0,1000)
ConverttoYV12()


just wma... and producing a wmv file with both audio and video muxed...

I tried with the following script:

Colorbars()
Trim(0,1000)
ConverttoYV12()
ConvertAudioTo16bit()

It produced a working asf file with audio, but I had to create an encoder profile and specify it with the -pr switch.

[Midzuki]

I've made a small change to completely ignore audio when the output is .vc1 or .raw. New version is here (http://www.mediafire.com/file/nnmajnyzzng/AVS2ASF.zip).

Thanks! I will test it A.S.A.P.

[buzzqw]

It produced a working asf file with audio, but I had to create an encoder profile and specify it with the -pr switch.
yes, but would be very very handy if avs2asf will accept the same audio encoder parameters as wmcmd.vbs

BHH

[TinTime]

I see what you mean. To be honest I'm not really interested in the audio side of things - my goal was really just to get the video encoding into a more stable state so I could start trying out different options without it failing inexplicably.

That said, I might have a look and see if I can make any sense of the audio code. I guess my first task will be to see if I can get it to work without specifying an external encoder profile. If I can then perhaps adding the kind of audio options you want won't be too hard. But don't hold your breath :)

By the way, if anyone wants the source code I'm using I can post that too. In fact I'll include it with any future updates.

[benwaggoner]

Well, I think it wouldn't be terribly difficult to call libaften and lame_enc. :) Anyway, WMA is not suitable for the containers MP4 and M2TS ;)
We've got WMA working just fine in MP4 for Smooth Streaming, FWIW.

@ benwaggoner: it seems you missed my whole point.
Even when I use "insane" settings for VC-1 encoding,
duplicated frames will still happen.
There's something wrong in your pipeline or settings then. We've exaustively tested that SDK with lots of long-form content, and simply have never seen it inject duplicated frames. If you're in Elementary Stream Sequence Header mode with proper settings, if it has to drop a frame due to VBV restrictions, it'll insert a proper skip_frame tag.

And when the ASF output is chosen,
the compressor may generate less frames than the framecount of the source file (1095 frames instead of 1100 frames, for example).
The container is a muxing post-process and so won't have any impact on the elementary bitstream.

I serious doubt the WMV~encoders didn't behave so
before the "VC-1 Era".
I think all the frame dropping issues had been fixed by the 2006 Format SDK 11, which is a generation before the VC-1 encoder SDK.

[Midzuki]

TinTime: now the program closes properly after encoding a .vc1 file, but it still "stalls" after creating a .ASF. I would like to know why this bad behavior appears not only in the WMV9 Sample Encoder, but also in ffmpeg2theora, WMVmuxer, and wm8eutil. :)

As for the WMA support, that's not relevant to me. If/when I want to create a WMA file, I can use Winamp, or WMNicEnc (yes, the GUI of Windows Media Encoder is a "mess", I really can't stand it. :p ). For muxing, AsfBin or WMVMuxer can do the job, fast and well.

benwaggoner wrote:
We've got WMA working just fine in MP4 for Smooth Streaming, FWIW.

Good! Now all that we need is a MP4 muxer and a MP4 splitter that support WMA in MP4. (BTW, what about VC-1 in MP4?)

If you're in Elementary Stream Sequence Header mode with proper settings, if it has to drop a frame due to VBV restrictions, it'll insert a proper skip_frame tag.

Wouldn't it be possible to rewrite the encoder in order to avoid the frame dropping completely ??? If I am not recording a live video stream, what is the point of dropping frames during the compression process? :confused:
The container is a muxing post-process and so won't have any impact on the elementary bitstream.

Perhaps...

The Windows Media Audio and Video codecs have been developed
to use the properties of the Advanced Systems Format (ASF),
which is the container used by Windows Media.

One of the ways in which the Windows Media Video codecs
produce video content for low bit rates is
by dropping duplicate frames.

{
source: msdn.microsoft.com
}

[TinTime]

TinTime: now the program closes properly after encoding a .vc1 file, but it still "stalls" after creating a .ASF. I would like to know why this bad behavior appears not only in the WMV9 Sample Encoder, but also in ffmpeg2theora, WMVmuxer, and wm8eutil. :)

I'm not sure why this should be - all I can say is that this doesn't happen on my pc. The last thing the asf writer does is update the indexing, etc. so for a large file it may appear to stall after it's finished encoding whilst it's doing this. Perhaps something else like explorer is trying to access it when this is happening? Pure guesswork on my part though.

Anyhow, there's a new version (http://www.mediafire.com/file/ztl0tdilg4n/AVS2ASF.zip), with the tiniest of changes. It just skips a bit of asf code when writing .vc1 output, but functionally there's no difference. The main reason for recompiling is that I've updated to the Win 7 SDK as per Kurtnoise's suggestion. Also I wanted to include the source with the download.

[Midzuki]

I'm not sure why this should be - all I can say is that this doesn't happen on my pc.
.
.
.
Perhaps something else like explorer is trying to access it when this is happening?

Probably it's because your PC is full of SDKs, compilers, poorly-designed runtime environments ;) , etc., — whereas my obsolete Pentium-4 isn't. :D

Anyhow, there's a new version (http://www.mediafire.com/file/ztl0tdilg4n/AVS2ASF.zip), with the tiniest of changes. It just skips a bit of asf code when writing .vc1 output, but functionally there's no difference. The main reason for recompiling is that I've updated to the Win 7 SDK as per Kurtnoise's suggestion. Also I wanted to include the source with the download.

Thanks.

[benwaggoner]

As for the WMA support, that's not relevant to me. If/when I want to create a WMA file, I can use Winamp, or WMNicEnc (yes, the GUI of Windows Media Encoder is a "mess", I really can't stand it. :p ). For muxing, AsfBin or WMVMuxer can do the job, fast and well.
Try the free Expression Encoder 3, or even WMP. I'm not a fan of the WME GUI either.

benwaggoner wrote:
Good! Now all that we need is a MP4 muxer and a MP4 splitter that support WMA in MP4. (BTW, what about VC-1 in MP4?)
MPEG has an official mapping spec for VC-1 in MPEG-4, which we used for the fragmented MPEG-4 Smooth Streaming implementation.

Wouldn't it be possible to rewrite the encoder in order to avoid the frame dropping completely ??? If I am not recording a live video stream, what is the point of dropping frames during the compression process? :confused:
I think you'll only see frame dropping in CBR modes, in fact. The main time I see significant frame dropping is if a max QP per frame is set too low. Leave it at the default max of 31.

If there's a mismatch between source and output frame rates, that could also yield duplicate frames due to judder, although that's rarer.

One of the ways in which the Windows Media Video codecs produce video content for low bit rates is
by dropping duplicate frames.

{
source: msdn.microsoft.com
}
This was true of older versions of the Format SDK, but was removed for VBR as of Format SDK 11, which the VC-1 Encoder SDK is derived from.

[Midzuki]

Try the free Expression Encoder 3

When I buy a new PC and install Windows Vista or Windows Seven on it, surely. :)
I'm not a fan of the .NET FrameworkS.

I think you'll only see frame dropping in CBR modes, in fact. The main time I see significant frame dropping is if a max QP per frame is set too low. Leave it at the default max of 31.

Sorry, but it seems that neither AVI2ASF, nor AVS2ASF are aware of that. :(

This was true of older versions of the Format SDK, but was removed for VBR as of Format SDK 11, which the VC-1 Encoder SDK is derived from.

As I said above, it seems that neither AVI2ASF, nor AVS2ASF are aware of that.

BTW: dropping or duplicating frames should have never been present in any of the WMV9 implementations, to begin with. :p To drop and to duplicate frames is a decision that should always be taken by the capturing/streaming application, NOT by the codec itself. At least XviD and Mencoder give to the user the freedom to choose, whereas the Windows Media Video encoders give me no choice at all. :(

[Inspector.Gadget]

Try the free Expression Encoder 3, or even WMP. I'm not a fan of the WME GUI either.

Wow, EE3 free is sharp! I've been playing around with it today and there's only one thing I can't figure out: how to get the output of input from Avisynth video and WAV audio to go to the same file. What I need to do is point the encoded video and audio to the same file, rather than separate ASF containers. I can work around this in Avisynth with AudioDub(), but that's annoying. Where's the switch I'm missing? Thanks.

[benwaggoner]

Wow, EE3 free is sharp! I've been playing around with it today and there's only one thing I can't figure out: how to get the output of input from Avisynth video and WAV audio to go to the same file. What I need to do is point the encoded video and audio to the same file, rather than separate ASF containers. I can work around this in Avisynth with AudioDub(), but that's annoying. Where's the switch I'm missing? Thanks.
Go into the "Enhance" tab and select your WAV source as the Audio Overlay.

[benwaggoner]

When I buy a new PC and install Windows Vista or Windows Seven on it, surely. :)
I'm not a fan of the .NET FrameworkS.
It's supported back to XP. If you have an objection to installing .NET, I can't help you though :).

BTW: dropping or duplicating frames should have never been present in any of the WMV9 implementations, to begin with. :p To drop and to duplicate frames is a decision that should always be taken by the capturing/streaming application, NOT by the codec itself. At least XviD and Mencoder give to the user the freedom to choose, whereas the Windows Media Video encoders give me no choice at all. :(
It's always been under programmatic control, via the "Quality" slider. A value of 0 sets the Max QP to 31, and eliminates CBR frame dropping in all but the most extreme cases as long as your using VC-1 Encoder SDK or at least the Format SDK 11 (WMP 11/Vista).

FSDK 11 and later should not drop frames in VBR encoding under any normal circumstances (I'm not really sure what happens if you set Min QP to 1).

Also, if you're using Advanced Profile properly configured, even in CBR mode it would introduce an explicit frame_repeat tag, not just extend the duration of the previous frame.

[Inspector.Gadget]

Go into the "Enhance" tab and select your WAV source as the Audio Overlay.

Thanks!

[Midzuki]

benwaggoner wrote:

It's always been under programmatic control, via the "Quality" slider.

I would prefer a checkbox with the option "Never drop frames". So the encoder would have to either increase bitrate and quality, or generate artifacts.

<!-- I'm leaving this thread behind -->

[benwaggoner]

benwaggoner wrote:



I would prefer a checkbox with the option "Never drop frames". So the encoder would have to either increase bitrate and quality, or generate artifacts.

<!-- I'm leaving this thread behind -->
Well, I'd say that respecting the VBV is the first order of business; a non-compliant file is worse than a missing frame. That said, I certainly think older versions of Windows Media, like other streaming-era codecs, dropped frames too much with default settings.

But it really hasn't been a problem in any of the implementations of the last 3 years.

Just using 2-pass VBR where the peak=average probably would give you this behavior I believe. And having non-pathological bitrates is always a big help :).

[Jerry_Sm@rt]

@TinTime
mediafire is blocked here,would you make a mirror download?

[Biggiesized]

@TinTime
mediafire is blocked here,would you make a mirror download?
What mirrors aren't blocked?

[TinTime]

@TinTime
mediafire is blocked here,would you make a mirror download?

Try this:

http://www.tintime.talktalk.net/AVS2ASF.zip

[Jerry_Sm@rt]

thanx.

[filler56789]

@comatose

Download this updated version. All the command line switches work with this fixed version.

The mediafire link is dead, does anybody have a copy of that file?

[Reel.Deel]

Try this one (http://forum.doom9.org/showpost.php?p=1336993&postcount=211).

[filler56789]

Try this one (http://forum.doom9.org/showpost.php?p=1336993&postcount=211).

Thanks for trying to help, but that's not what I am looking for (Tintime's AVS2ASF encodes to Advanced Profile only).

[filler56789]

has there ever been made a GUI for this in the meantime?

Yes, BUT...
Someone was asking for a GUI for this encoder - here's one:

http://extranet.spitzinc.com/download/public/misc/VC123.zip <= 404 ERROR :(

This is a Visual Studio '08 project that uses MFC. It's basically AVI2ASF plus a big dumb dialog for all the parameters. This is a work in progress on the way to becoming a fairly narrow-purpose in-house tool, but others might find it useful for experimenting with the VC-1 encoder. I've left a pre-built executable in the Release folder for anyone without an IDE.

Read UsageNotes.txt

I'd be happy to entertain suggestions for improvement, but can't promise to implement anything . . .

Anyway, Nic's AVS2ASF is the least usable one. :( TinTime's mod is quite usable, however it doesn't support Simple and Main profiles, and NorthPole's builds are not available anymore. :(

FWIW,
TinTime
Registered User

Last Activity: 10th September 2011 23:28

NorthPole
Registered User

Last Activity: 6th September 2011 21:29

[DragonZord]

I've been trying to get a VC-1 encoder set up for VirtualDub. It's proving difficult.

Benwaggoner over on Doom10 suggested (http://doom10.org/index.php?topic=2554.msg11655#msg11655) the AVS2ASF solution, but all I can get is an Error -20. I have Expression Encoder v4 installed but no information on how to set-up AVS2ASF. I'd guess that it needs some DLLs from Expression Encoder? Is my version even compatible?

Is this an appropriate solution? If so, how do you set it up prior to encoding?

Unfortunately without stdin on any of the VC-1 encoders I'll have to use a temp file. VirtualDub's limited implementation of external encoders doesn't favour this approach, but I have it working, even if it makes a joke of the progress bar during the encode process.

Help very much appreciated. Thanks :-)

[DragonZord]

Why? Just curious.

Because not every use for digital video is related to leisure viewing. At home I like my videos in mp4(x264+aac). At work, we use wmv(wvc1) for compatibility reasons. The sort of videos we produce are used in presentations and sent around the world. WMV is about the only thing that isn't routinely answered with complaints, and given that we use Expression Encoder, our stuff has historically been in VC-1 which actually does a great job. WMVs from ffmpeg do get returned sometimes which is why I'm keen to keep an official VC-1 implementation.

Part two of the story is that I like VirtualDub. It's not perfect and personally I think the way they've implemented external encoders is limited, awkward and badly integrated, but that's off topic. I've nothing against moving to AviSynth one day, except that I find with text-based interfaces you often forget how to do things unless you use them all the time. And that's very inefficient. Case and point, I could use AviSynth very well in 2004. All I can do now is DirectShowInput().

So the crux is that it'd be very nice if when I've finished my video I can just make it into a wmv(wvc1) by File > Export. I am obviously putting too much time into this, but I figured that I'd have solved it by now and in the process published a step-by-step solution online in a nice concise forum post. Something the Internet is missing at the moment.

[DragonZord]

What do you mean by this? It reads like as if you already have solved it and already published such a post?

I didn't start here. I started about 10 days ago but couldn't find my old Doom9 login. And no it doesn't read like that: it's past conditional (http://www.edufind.com/english/grammar/if_conditional1.php).

In VirtualDub you probably could export your finished video in a lossless intermediate format and then import and encode that in Expression Encoder 4 SP2 with VC-1?

Yep, that's what I do. I figured the collective wit of mankind might be able to eliminate that step though.

[benwaggoner]

Yep, that's what I do. I figured the collective wit of mankind might be able to eliminate that step though.
EEv4 supports live encoding. So if you could hack VirtualDub to output a frame server as a DirectShow video source, then that could work.

However, I'm having trouble coming up with a problem that would be a good way to solve. EEv4 SP2's internal preprocessing is generally as good as VirtualDub, and it can also read AVS directly as long as the per-frame render time isn't so slow that the encoder times out.

In those cases, I'll just dump to a LAGS 4:2:0 AVI, and EE will take that perfectly and pass on the pixels to the encoder as is.

More broadly, what do you need really good VC-1 for in 2013? I see its value today most in its being broadly compatible with old Windows managed desktops in enterprise environments.

[DragonZord]

...what do you need really good VC-1 for in 2013? I see its value today most in its being broadly compatible with old Windows managed desktops in enterprise environments.

Yep. I think I said that didn't I?

[benwaggoner]

Yep. I think I said that didn't I?
Okay. Could you just use Expression Encoder, then? It's a very nice tool for doing encoding for typical enterprise scenarios, and certainly takes a lot less time per file than anything involving VirtualDub or a command line.

You get nice features like parallel encode of multiple bitrates of a single source, automatic settings configuration based on interlaced/progressive source, aspect ratio, and frame rate, very nice scaling algorithms, a nice audio normalization filter, etcetera.

For most enterprise scenarios I can think of, EEv4 will give pretty much the optimal possible output for VC-1. And since it has the final Enterprise SDK version, it has the quality and speed improvements from that which were never added to the Professional version.

[DragonZord]

Okay. Could you just use Expression Encoder, then? It's a very nice tool...

That's an option, yes. Personally I don't think it is a nice tool though. Just now it failed to encode something and now it won't import video anymore. Just another painful morning with this tool. I've deleted the user settings (etc) and re-installed and still it's broken. Joy.

Maybe I will focus my attention on wmcmd.vbs or see if I can bribe someone to add stdin. I'm not really bothered that it's out of date as it is at least reliable and can make decent videos.

[benwaggoner]

That's an option, yes. Personally I don't think it is a nice tool though. Just now it failed to encode something and now it won't import video anymore. Just another painful morning with this tool. I've deleted the user settings (etc) and re-installed and still it's broken. Joy.
SP2 installed?

Also, by default EE will use DXVA to decode supported formats, which could possibly cause issues with funky bitstreams. You can control which decoder does what in

Maybe I will focus my attention on wmcmd.vbs or see if I can bribe someone to add stdin. I'm not really bothered that it's out of date as it is at least reliable and can make decent videos.
If you're on at least Windows 7, you'll get the 8-way threading, 30% speedup improvement, and "swirly fix" from the final Enterprise release. Plus a better low bitrate DQuant. No Lookahead Rate Control, though. But on the whole that is probably as good or better as the existing Pro library for most scenarios.

An app that calls the OS Windows Media encoding functionality, using a PropertyBucket to control parameters, could be a lot easier to write. And getting stdin to work with DirectShow should be pretty trivial. You could probably prototype the whole thing in GraphEdit if you wanted.

[DarrellS]

Hey DragonZord, did you get anything working in Virtualdub?

I've got WMA-STD, WMA-LSL and WMA-PRO working with WMAEncode_0.2.9b but I cannot get 5.1 with any encoder I've tried. The cscript/wmcmd.vbs option isn't working for me at all with audio or video. Any help would be appreciated.

Here are the commands for WMA using wmaencode.exe...

{
"description": "VirtualDub external encoder profile collection",
"externalEncoders": {
"sets": {
"WMA-STD": {
"videoEncoder": "",
"audioEncoder": "WMA-STD",
"multiplexer": "",
"description": "wma",
"extension": "wma",
"processPartial": true,
"useOutputAsTemp": false
},
"WMA-PRO": {
"videoEncoder": "",
"audioEncoder": "WMA-PRO",
"multiplexer": "",
"description": "wma",
"extension": "wma",
"processPartial": false,
"useOutputAsTemp": false
},
"WMA-LSL": {
"videoEncoder": "",
"audioEncoder": "WMA-LSL",
"multiplexer": "",
"description": "wma",
"extension": "wma",
"processPartial": false,
"useOutputAsTemp": false
},
"profiles": {
"ASFmux": {
"name": "ASFmux",
"program": "ffmpeg-20130624-git\\ffmpeg.exe",
"commandArguments": "-i \"%(tempvideofile)\" -i \"%(tempaudiofile)\" -vcodec copy -acodec copy -f wmv -y \"%(outputname)\"",
"outputFilename": "%(outputname).audio",
"type": 2,
"inputFormat": 0,
"checkReturnCode": true,
"logStdout": true,
"logStderr": true,
"bypassCompression": false,
"predeleteOutputFile": true
},
"WMA-STD": {
"name": "WMA-STD",
"program": "WMAEncode.exe",
"commandArguments": "-i \"-\" -c std -m cbr \"%(tempaudiofile)\"",
"outputFilename": "%(outputname).wma",
"type": 1,
"inputFormat": 1,
"checkReturnCode": true,
"logStdout": true,
"logStderr": true,
"bypassCompression": true,
"predeleteOutputFile": true
},
"WMA-PRO": {
"name": "WMA-PRO",
"program": "WMAEncode.exe",
"commandArguments": "-i \"-\" -c pro -m cbr \"%(tempaudiofile)\"",
"outputFilename": "%(outputname).wma",
"type": 1,
"inputFormat": 1,
"checkReturnCode": true,
"logStdout": true,
"logStderr": true,
"bypassCompression": true,
"predeleteOutputFile": true
},
"WMA-LSL": {
"name": "WMA-LSL",
"program": "WMAEncode.exe",
"commandArguments": "-i \"-\" -c lsl \"%(tempaudiofile)\"",
"outputFilename": "%(outputname).wma",
"type": 1,
"inputFormat": 1,
"checkReturnCode": true,
"logStdout": true,
"logStderr": true,
"bypassCompression": true,
"predeleteOutputFile": true
},
}
}
}
}


Not sure if ASFmux will work since I can't get any wmv encoders working. ffmpeg or wmcmd.

[raffriff42]

>>In VirtualDub you probably could export your finished video in a
>>lossless intermediate format and then import and encode
>>that in Expression Encoder 4 SP2 with VC-1?
Yep, that's what I do. I figured the collective wit of mankind might be able to eliminate that step though.You want to eliminate the intermediate file, or eliminate VirtualDub entirely? To eliminate VirtualDub, skip to step 4 Frame serve from VirtualDub.
Save signpost as .avi, not .vdr as documented.
Do not open signpost .avi in MS Expression 4.
Create Avisynth wrapper script.
Open script in MS Expression 4.
Encode. Share and enjoy. :)Tested with VirtualDub 1.10.3, Avisynth 2.6, MS Expression Encoder 4 SP 1

[DarrellS]

I think the idea is to do it all in Virtualdub with the external encoder feature using command line encoders and multiplexer without having to use two or three other programs to do the job.

ffmpeg supports wmv1 and wmv2 but does not support wmv3 or vc-1 encoding.
cmdvbs doesn't support raw input from Virtualdub so we can't do audio or video
asfbin will not mux the wmv2 and wma2 audio from Virtualdub
ffmpeg will mux the files but the files won't play in wmp or mpc-hc (the files will seek but not play). They do play in VLC player.

If the guy that created wmaenc.exe would've created a wmvenc.exe then we'd be in business since we are able to create windows media audio files with his command line encoder. We just can't create 5.1 audio files (not sure why since it uses the pro codec).

I have no desire to encode wmv for myself but I am interested in getting as many command line encoders as possible working with the external encoder feature of Virtualdub. It looked like DragonZord was really close but as long as the cmdvbs option doesn't support stdin then I don't think it's possible and I don't think anyone is going to create a wmv command line encoder anytime soon (or ever). ffmpeg could implement a wmv9 encoder and make it easy for everyone since MS released wwv9vcm years ago but I guess they're afraid of a lawsuit. I know someone was working on a vc1 encoder for ffmpeg years ago but the project was dropped.

[filler56789]

I was bored, so I archived TinTime's mod of AVS2ASF @ GitHub:

https://github.com/filler56789/AVS2ASF-2