Virtualdub: Combining audio and video files in workflow.

[Cronjob]

I'm a Unix guy who spends his life at the command line doing enterprise-y stuff, so I'm a newb at recording and editing media. I've been educating myself, but am running into a hurdle in my workflow. I've googled endlessly and can find no solution. I am frustrated, because I believe it's probably something obvious and it is my lack of terminology or understanding which is the real hurdle, here. Either way, I'm hoping for some guidance and insight here to nudge me along.

The workflow I'm trying to use is as follows:

1) Record content with Dxtory, using the Lagarith video codec and setting audio to PCM or AC3.

2) Open resulting file in Virtualdub.

3) Use Virtualdub to encode to H.264.

This is where I run into a problem. After encoding, I have a separate video and audio file. I can't figure out how to make the output merge both of these or, if necessary (though perhaps unnecessarily extra) how to then take these two files and run them through Virtualdub *again* to merge them. I have tried telling it to convert the audio. I've tried telling it not to convert the audio. I've tried taking the encoded video and loading it back into Virtualdub and then telling it to use the audio file from the previous encoding and load that as its audio source. No luck.


What am I missing or misunderstanding? I've gone through so many discussions and guides and youtube tutorials I've found via google and they always stop at the "SAVE AS AVI" step and don't then explain what to do with the two separate files at the end. Am I wrong in understanding that I should be able to accomplish this in Virtualdub? Or am I supposed to go get yet *another* application to merge these?

Any help is surely appreciated.

Update: Wow, after an entire day, I was able to get it to output into one file using the method I'm pretty sure I've done a dozen times, now (told it to use direct stream copy audio). Of course, now it's just stuttery and wildly out of sync, but I guess that's a common problem, anyway.

[Guest34343]

You're a command line guy so why wouldn't you use the CLI x264 encoder? Encoding to AVC through VirtualDub is considered a hack by most people.

Encode your video using an Avisynth script and x264.exe. Then mux your encoded video and audio into an MKV file using mkvtoolnix.

[Cronjob]

Thanks for the response. Here's an overly detailed update, for anyone who is remotely interested.


When going from nearly absolute zero, the sheer amount of information to absorb out there (and all the rabbit-holes you travel down as one thing leads to the next) is totally overwhelming. It wasn't until shortly after I posted that I discovered that VFW is essentially legacy stuff and that in addition to other versions of x264 codec, there was a CLI version (which obviously made things much easier).

I did figure out what I was doing wrong in VirtualDub. It was a misunderstanding of the menu system for the Audio Encoding section.

Ultimately, I also discovered that MeGUI was more than just an interface to the x264 codec and provided a number of rather convenient facilities (I had dismissed it when I saw it, before, not realizing exactly what it was). So that is the direction I've gone. My workflow now seems pretty decent for my limited needs:

1) Record content with Dxtory, using Lagarith and PCM 48khz 16bit Stereo
:::::1a) Recording is from a 2560x1600 display running at native resolution, converted into 1152x720 via Dxtory to maintain 720p and proper 16:10 aspect ratio. I'm using RGB24 color space, but somehow my encoding processes forces a conversion to YV12 *no matter what*, which I'd rather not do -- but if I have no choice, then I'll change Dxtory to use YV12, too. Might as well save a step.

2) Convert to x.264 in MP4 via MeGUI and audio to 448 ABR FAAC.
:::::2a) Using 2-pass 10240 bitrate with High AVC Profile and Very Slow preset. CABAC enabled. GOP calculation FPS based, B-Frames set to 3, adaptive B-Frames set to Optimal, B-Pyramid disabled, 5 reference frames, 40 extra I-Frames, 10 (QP-RD) subpixel refinement, and Trellis set to 'always'. I'm iffy on a few of these settings (such as Trellis, which I understand is demanding).
:::::2b) One weird thing (though probably not, if you understand what I must not) is that even though I'm using 448 ABR on my audio stream, it seems to come out as 165. I assume the 165 is *per channel*? If not, then I can't understand why it'd go from uncompressed PCM to 165 (even 165x2 seems low) unless the amount of audio data I'm starting with is just so ridiculously light (it's the audio to a game). I'm not currently certainly exactly how to investigate that further though, so...

3) Merge/split files as necessary afterward to maintain desired length/size constraints.


The result is an 1152x720 file that looks pretty great, even blown up to 2560x1600 and for about 4.5gb/hr.

The only downside is the encoding time, which runs about 7x to 8x the content time, even with priority set to HIGH, but that seems to be due to multi-threaded support, perhaps, at some step in the encoding process (I'm watching a first-pass right now, which is using less than 15% CPU and 25% disk I/O). Even pulling the preset all the way back Medium and reducing a couple likely heavy options in configuration doesn't change that.

(By the way, I have nothing against MKV, but it seems MP4 is a more accessible container for play on wider variety of devices and, long term, will probably have more support than MKV will in a decade or longer -- but perhaps I am entirely wrong about this?). At any rate, it's not likely that I'll need a complex container. I only need the video and then the audio stream. POSSIBLY in some cases, a second and (rarely) a third stream containing commentary.

Anyway, I've learned a ton in the last week and change and what once seemed monumentally confusing and involved is slowly seeping in. I honestly never fathomed just how utterly complex the world of video encoding was. Fascinating!