To separate or not to separate that is the questio

PeterDuke wrote on 5/6/2014, 11:01 PM
I thought the conventional wisdom when using DVD Architect was that the audio and video had to be in separate files, or the audio would be (re-)compressed. On pages 196 and 197 of the DVDA 6 manual, for instance, it says “You’ll need to render your audio stream separately …”.

This practice originated before the arrival Blu-ray discs. To make a DVD, you would normally use the MainConcept MPEG-2 encoder in Vegas for the video, but this will only create an MPA (MP2 or MPEG-1 layer 2) audio stream, whereas DVDA requires linear PCM or AC-3 (stereo or 5.1). Hence you need to render your audio stream separately.

Then came Blu-ray discs. If I render my AVCHD files using the Sony AVC encoder to make an m2ts file with the audio included, I receive a message that the audio would be compressed (not re-compressed). If I use a separate audio file rendered using the Dolby Digital AC-3 encoder there is no such message. Hence, once again, you need to render your audio stream separately.

From my experience above, and from countless comments made by others in these forums, I thought that the inflexible rule was that the audio and video should always be in separate files, whether making a DVD or BD.

In another thread concerned with remaking a DVD http://www.sonycreativesoftware.com/forums/ShowMessage.asp?MessageID=894084&Replies=25 I made this statement:

Another way to [remake a DVD] might be to open DVDArch and create a new DVD project as appropriate. Drag VTS_01_0.IFO from your DVD to the menu in DVDArch. This will load the audio and video plus chapter markers. Edit the markers as you like. You will need to create a new menu from scratch, however.

Due to DVDArch's stupidity, it will re-encode the audio because it is included with the video and not in a separate file. You could prevent this by extracting and concatenating the VOB files, for example with VOB2MPG, and then demuxing with say TsMuxerGUI to produce an audio-only file, and using this to replace the audio in your project.

And then:

I just went through the exercise I have been suggesting, using a recently made DVD.

When I go to make the DVD, DVDArch says that it will recompress the video AND compress (not recompress) the audio. I set it going and it was obviously recompressing the video because it said so and was taking many minutes in the process. I had thought that it would not recompress a video it had previously accepted without recompression, but wrong again!

So this procedure is suboptimal.

And finally:

I just made a DVD project using the MPG file from a DVD extracted using VOB2MPG. It contained BOTH the audio and the video. I inserted some markers and made a scene menu, and then created a DVD (folders only).

IT DID NOT RECOMPRESS EITHER VIDEO OR AUDIO!

As a further exercise, I then took the MPG and AC3 files created by Vegas and muxed them to a TS file using TsMuxer, and then converted it to an MPG file (PS) using VideoReDo. A DVD made from this file took 26 secs and THE AUDIO WAS NOT COMPRESSED.

Thus, for a DVD, if the audio is included with the video and both are compliant, then life is sweet. But this did not seem to be true for a BD…

I therefore made a simple AVCHD project in Vegas 9c and rendered out the timeline containing hundreds of unaltered small clips. (The video of all but three clips was smart rendered.) I created a video+audio file, a video-only file and an audio-only file. The audio was 5.1 surround. The timeline was about 43 mins long.

The render times were as follows:
Video+audio 9:29
Video 5:27
Audio 8:08
(A smart render of audio in Vegas would be useful!)

I then made a simple BD project in DVDA Pro 5.2 with a single menu page and no chapters and created a BD ISO file. The render and prepare times were:

Combined video & audio file 8:26, 3:54. (I was advised that the audio would be compressed.)
Separate video and audio 0:36, 3:28

I then demuxed the combined file into elementary video and audio streams using TsMuxer and processed once more. When loading the raw video stream, DVDA took quite a while to digest it, compared to the video in the m2ts container. I then added the audio and created a BD ISO file once more. I was advised that the audio would be compressed EVEN THOUGH IT WAS IN A SEPARATE FILE. The render and prepare times were 7:50, 3:53 respectively.

I then took the first two ISO files, mounted them with Virtual CloneDrive and prepared a new ISO file from each main M2TS file. The render and prepare times were virtually the same, about 2:55, 4:03 AND THE AUDIO WAS NOT COMPRESSED EVEN THOUGH THE VIDEO AND AUDIO WERE IN THE SAME FILE.

Finally, I took the video+audio M2TS file and the audio AC3 file from Vegas renders and remuxed them to an M2TS file with only the latter audio stream. During creation of the ISO file, THE AUDIO WAS NOT COMPRESSED. Render and prepare times were 1:08, 5:15.

The render times when the audio was not compressed ranged from 0:36 to 2:55, and when the audio was compressed, from 7:50 to 8:26.

The prepare times over all tests, ranged from 3:28 to 5:15. The time was longer when the video and audio were combined in the one file.

It would therefore appear that the video and audio streams may be combined into the one file provided that they are “compliant” (read: “acceptable to DVDA”). The Sony AVC encoder does not apparently produce “compliant” audio.

Comments

No comments yet - be the first to write a comment...