Questions on video/audio streams

tfc wrote on 9/21/2004, 9:07 PM
Let's say I have a project and I wish to render it to MPEG-2. My ultimate goal is to separate both the video and audio files. Under the "Render As" menu, I have the option of unchecking the box entitled "include video or audio stream" under it's respective tab. If I uncheck one of these, say, for example, the audio stream, the video file only, as I understand it, gets rendered to MPEG-2. The same is true for the video side as well. Now, as an alternate technique, let's say that instead of rendering this way, I go to the last tab on the "Render As" menu, entitled, "System" and I check the box entitled, "save as separate elementary streams". This will render the file as both an .m2v and .mpa file extension. What is the difference between these two scenarios? Aren't they really accomplishing the same thing? When one uses the first scenario and unchecks the "include audio stream", for example, does the actual final rendered result actually have an audio stream, but is it just muted? In other words, does the render create some sort of "dummy" MPEG-2 audio file as some kind of blank placeholder, or is it really doing the same thing as the 2nd scenario and actually only creating just an .m2v file. Can somebody explain these two scenarios and what they mean? When would I want to use the first technique described above vs. the separate "elementary streams" technique? Which one would be the better way to accomplish a project with multiple audio tracks for the final DVD, for example? Does the way one separates the video and audio files via either of these methods have an impact at all on how DVD Architect will read/ react/ or deal with these files? Somebody, please school me on this! Thanks!!!

Comments

ScottW wrote on 9/22/2004, 5:44 AM
Since it's DVDA the answer is simple. Never render your audio as part of your video. DVDA wan'ts the audio and video as seperate files - if you hand it a mutiplexed stream, then it will simply re-render (which is a waste of time and can impact quality).

From the list of render templates, select the DVDA NTSC (or PAL if you're in PALland) video template. Then as a seperate step (or using one of the scripts found on VASST) render the audio as an AC3 file. Don't render your audio as mpeg2 - very few players support this format.

--Scott
tfc wrote on 9/22/2004, 11:56 AM
ScottW: "From the list of render templates, select the DVDA NTSC (or PAL if you're in PALland) video template. Then as a seperate step (or using one of the scripts found on VASST) render the audio as an AC3 file. Don't render your audio as mpeg2 - very few players support this format."

True, mpeg-2 audio, is not in the DVDA spec, I believe DVDA won't even burn the disc, unless it converts the audio to AC-3, or PCM audio first.

But, does anybody know the difference between these two above aforementioned techniques. From a technical point of view, are they the same or not? If not, what are the technical differences?
Chanimal wrote on 9/22/2004, 12:16 PM
I have a similar question.

I've just rendered the final video as an Mpeg2 with video and audio and let DVDA convert it to AC3 for me. It doesn't seem to take much time (so it is obviously not re-rendering the video) and it allows me to preview the Mpeg2 prior to burning a DVD (plus I have it for desktop use). I can see the advantage of splitting out audio if I use surround sound, but I don't the much otherwise.

Is there really any loss in video quality?

***************
Ted Finch
Chanimal.com

Windows 11 Pro, i9 (10850k - 20 logical cores), Corsair water-cooled, MSI Gaming Plus motherboard, 64 GB Corsair RAM, 4 Samsung Pro SSD drives (1 GB, 2 GB, 2 GB and 4 GB), AMD video Radeo RX 580, 4 Dell HD monitors.Canon 80d DSL camera with Rhode mic, Zoom H4 mic. Vegas Pro 21 Edit (user since Vegas 2.0), Camtasia (latest), JumpBacks, etc.

tfc wrote on 9/22/2004, 12:35 PM
Chanimal: "I've just rendered the final video as an Mpeg2 with video and audio and let DVDA convert it to AC3 for me. It doesn't seem to take much time (so it is obviously not re-rendering the video) and it allows me to preview the Mpeg2 prior to burning a DVD (plus I have it for desktop use). I can see the advantage of splitting out audio if I use surround sound, but I don't the much otherwise.

Is there really any loss in video quality?"

In case you didn't know, you can definitively tell whether DVDA is recompressing audio or video. When you go to prepare and/or burn your DVD, go to the optimize button and hit that. On the left side is a list of all the files. At the top there are two little icons. The one on the left is the video portion and the one on the right is the audio portion, denoted by a little filmstrip and music symbol respectively. A green check mark in either column means that that respective file is compliant, in other words does not need to be recompressed. You can also tell another way. If you put the cursor on that file in question, and go to the far right hand side, you can see all of the properties of that file which is highlighted. You can go under the video or audio column and look under the heading of "recompress". It will say either yes or no. This should match whether there is a green check mark or not in the other file column on the left hand side.

You ask is there a loss of video quality? The answer is no. It doesn't matter where the video file was rendered, it either is or is not MPEG-2 compliant. I think what you really meant to ask is about the audio. This is a question I have had also. That is, is it better to have Vegas render the audio directly to AC-3, rather than have it render it to MPEG-2 audio, and THEN have it rendered to AC-3 in DVDA. In other words, doesn't it make sense to compress the audio only once, going from the original .WAV file directly to AC-3, vs. going from .WAV to MPEG-2 audio, and THEN to AC-3. Is my logic correct here. Can anyone add any details?
Spot|DSE wrote on 9/22/2004, 12:53 PM
There are no technical differences between rendering the files separately or rendering using a batch script.
ScottW wrote on 9/22/2004, 12:55 PM
The difference is the type of streams created. In the case of rendering with the DVDA PAL/NTSC template, the encoder is going to create a system stream (aka program stream) - regardless of whether it has audio in it or not.

In fact you can have a system/program stream that contains just audio, just video or both.

In the second case you are creating an elementary stream. These streams can not have multiplexed data - they are either audio or video but never audio and video.
--Scott
ScottW wrote on 9/22/2004, 1:00 PM
MPEG-2, like AC3 is a lossy compression method. If you render your audio as MPEG-2, you've lost some of the info contained in the original source. Then by having DVDA re-render it as AC3 you've lost additional information.

Can you detect this? Maybe not, but still, why run the risk when you can just render as AC3 and avoid the possibility entirely.

--Scott
tfc wrote on 9/22/2004, 2:09 PM
Scott W:"The difference is the type of streams created. In the case of rendering with the DVDA PAL/NTSC template, the encoder is going to create a system stream (aka program stream) - regardless of whether it has audio in it or not.

In fact you can have a system/program stream that contains just audio, just video or both.

In the second case you are creating an elementary stream. These streams can not have multiplexed data - they are either audio or video but never audio and video."

Okay, thanks for the info! So what effect does this have on the quality and/or compatability of the steams? Is the difference really transparent to us. In other words, is it just another way to arrive at the same end result?
tfc wrote on 9/22/2004, 2:18 PM
ScottW:"MPEG-2, like AC3 is a lossy compression method. If you render your audio as MPEG-2, you've lost some of the info contained in the original source. Then by having DVDA re-render it as AC3 you've lost additional information.

Can you detect this? Maybe not, but still, why run the risk when you can just render as AC3 and avoid the possibility entirely."

This is exactly what I suspected. The only reason I ask this is that I had some problems doing this on a project. I rendered the video spearately from the audio. The video was rendered in V5 as MPEG-2 and the audio was rendered in V5 as 2 channel stereo AC-3. When I brought those both into DVDA and tried to lay down additional audio tracks for the video, DVDA "interpreted" the additional AC-3 audio tracks as a full audio and video track. In other words, the additional audio tracks somehow got "connected" to the video. They both had different file names, FWIW! Even though there was only one video track and 3 AC-3 audio tracks, DVDA added a video track to every AC-3 audio track! My disk usage jumped to like 9.4 GB! It should have been substantially less than 4.7! I have a feeling it was somehow related to the video/audio streams, but I can't figure out just why. This why I have asked these series of questions. Thanks everyone!
ScottW wrote on 9/22/2004, 4:14 PM
As I understand it, system streams also contain some additional timing information that's not present in elementary streams. Vegas will read system streams, but won't read elementary streams (go figure). Other than that the differences are transparent to mere mortals until you end up the wrong stream without a paddle (one of the reasons that it's handy to have TMPGenc around, since its mpeg utilities will let you multiplex and demultiplex with wild abandon).

For example, DVDA 1.0 won't touch elementary streams, though now 2.0 does. DVD Lab likes elementary streams rather than system streams, but will convert one to the other if needed.

Another fun irritant is the lack of consistancy in file naming. Usually system streams have a .mpg extension and elementary an .m2v, but sometimes you'll run into a .mpg that's really elementary.

--Scott