Authoring truncates video + weird muxing observation (pics)

[Brent212]

Okay, this is a two part question... the first sort of led to the second.

Here it is in a nutshell: I captured a.mpg with my hauppauge 250. The file played back nicely, with the audio and video perfectly in synch throughout. I scanned it with Womble's Mpeg-Vcr GOP Fixer, which checks for three types of errors, including timecode errors. No errors were detected. So far so good. I pop a.mpg into TMPGEnc DVD Author, leave it as is, without adding any chapters or anything and no menus. I then start the authoring process to create the VIDEO_TS folder and its files. The projected size is about 2.8 GB. When it's all done, I close TDA and check the size of the resulting VIDEO_TS folder. To my surprise, it's about 1.7 GB. I open it up in PowerDVD and begin playing. The video is only about 28 minutes, while the original a.mpg was about 45 minutes. The audio and video are in synch the whole time, it just gets cut off at 28 minutes.

No problem... I've never had this happen, but I know there are many strange errors that can occur in dvd authoring land, so I don't panic because there are many things I can do to avoid this problem. The first, and easiest is to simply demulitplex a.mpg and then re-multiplex it before sticking it into TDA. Sometimes that seems to fix these things for whatever reason.

So I demux a.mpg with Womble to a.mpv and a.mpa. I then re-mux a.mpv and a.mpa to a2.mpg. I scan a2.mpg with Womble's GOP Fixer and again get no errors. I play a2.mpg in media player classic, and it plays fine. Cool. I then repeat the authoring process in TDA, this time with a2.mpg instead of a.mpg. The VIDEO_TS folder comes out to be the predicted 2.8 GB, and the full video plays perfectly. Sweet.

Now I try something just out of curiosity. I compare a.mpg to a2.mpg. They should be identical, shouldn't they? I just demuxed and re-muxed a.mpg to get a2.mpg. I save a frame from a.mpg and save the same frame from a2.mpg. I then compare them, and the frame from a2.mpg is noticeably brighter than the frame from a.mpg, although the details and everything look exactly the same. Here are the two pics:

from a.mpg:


from a2.mpg


Weird, eh? The difference is totally noticeable while playing a.mpg and a2.mpg next to each other also.

So my questions are:

1) Does anyone have any idea why a.mpg wouldn't author correctly and a2.mpg did? What does demuxing and then re-muxing do to fix whatever the error was?

2) Why did the muliplexing process affect the picture? I wasn't aware that it would do anything like that. I know sometimes it will mess with the synch of the audio and video, but I didn't think it would actually alter a video stream. Could it somehow simply change a brightness or other color value (if such values exist)?

Thanks for listening, er... reading.
-Brent

[Brent212]

actually, now that I look at the pics again, I think the contrast was increased in a2.mpg.

The blacks are a little darker, but the lights are a little lighter. I sort of like it better, although altering the original cap worries me.

[Brent212]

hmmmm... I'm thinking an interesting test would be to demux a2.mpg (to a2.mpv and a2.mpa), and then compare a.mpv (the demuxed video stream from a.mpg) and to a2.mpv.

that's right... I am talking to myself.

[FulciLives]

that's right... I am talking to myself.

Do you happen to look like this when talking to yourself?



As for the problem here or difference encountered ... VERY ODD!!!

I would try to do this though:

Once you demutiplex there is no reason to then multiplex it back. TMPGEnc DVD Author will accept a m2v video file and mp2 audio file.

So try to demux then import the "elementry streams" and see what happens.

- John "FulciLives" Coleman

[Brent212]

Good idea. Although I always feel that if I give TDA two separate streams it's got a greater chance of getting the synch wrong. It's like handing a little kid a pair of scissors with the blades wide open... or maybe it's not like that at all. I just view TDA as somewhat incompetant at keeping the streams synched so I'd like to give it as much help as possible. I'm pretty sure TDA does muxing during the output process, so theoretically this shouldn't really create any more risk of synch problems. It just seems like superstition plays a big part in authoring, and any change in a working process will throw the whole universe out of wack (by 5 frames)

I guess if it turns out that a.mpv matches the video in a.mpg, then it means the "contrast" change happens during the re-muxing process, and this would be a way to avoid it. Right on.

PS - where'd you get my senior portrait?

[FulciLives]

I've done hundreds of projects with TMPGEnc DVD Author and always use elementry streams and have yet to have it cause any A/V sync problems.

So try it

- John "FulciLives" Coleman

P.S.
I guess another option would be to explore different programs that have a multiplex option.

[Brent212]

I've also used TDA quite a bit and rarely have problems, but every once in awhile...

Same with ULead MF 2. It seems to work fine most of the time, but then a random file with cause it to hickup.

Luckily I've yet to find a case where both TDA and MF have problems with the same file, so basically if one doesn't work the other will (knock on wood).

Do you use elementary streams on purpose, or do you just happen to have them separated at the point that you begin authoring? I've never heard a benefit of doing that, other than convenience.

[vhelp]

@ Brent212

I skimed through your post(s) and didn't see where you
mentioned what app you used to deMUX your A/V source

* Did you try other deMUX'ers (ie, TMPG's mpeg-tools) ??

* have you tried using another DVD authoring approach
.. for instance, ifoEDIT ??
.. ifoEdit is pretty easy to use, and IMO, it does not taint or
.. alter your mpeg (A/V de-MUXed) sources whatsoever.

Also, if you don't want to wadd through w/ TMPG's mpeg-tools,
(and you have ifoEdit) you can drog n drop your MPEG-2 file(s)
onto its icon and it will deMUX it on-the-fly for you into *.mp2 / *.mpa
source files

-vhelp

[trock]

It looks like a2.mpg has corrected the black level of a.mpg (or maybe vice versa). Different mpg encoders treat black levels differently but I've never heard of muxer/demuxers doing it.