Peak bitrate of the exported VOB files change radically

Question

I have tested with Bitrate Viewer my timeline's .m2v file, its peak bitrate is a bit under 8.7Mbits. The project has one subtitle track, and two ac3 audio tracks (2.0 track is 320kbps (Encore encodes the 2.0) and 5.1 is 448kbps). So, as making peak bitrate calculations the maximum bitrate should not exceed 9.8Mbits, right? Still, when I'm viewing the exported DVD Folder with Bitrate Viewer, VTS_01_1.VOB has a peak bitrate of 12723 bits. In that VOB the average bitrate is around 7500bits as it should and mostly max. 8.2Mbits. How is this possible that there is this single super peak in the VOB?
PS. Did another test, I changed the 2.0 AC3 track (which Encore encodes) to a previously AC3 encoded track my client provided me, now two of the VOBs have these super peaks as:
VTS_01_1.VOB - 11010 bits
VTS_01_5.VOB - 13486 bits
What is the logic?

Answer 1

Mick.
Whilst there is nothing to actually prevent you from using a 320kbps stereo AC3 stream at all, I was only referring to
i Encore's
encoding to AC3 stereo being 192kbps. AFAIK (I do not have a copy of the specs in front of me so cannot be 100% certain here) the specs merely state that the maximum bitrate for AC3 is 448kbps. The general consensus rate for stereo is 192, and Encore's Automatic transcoding uses this value. It is certainly possible to use the whole 448 in stereo, assuming your encoder is capable of this. I have even seen surround 5.1 streams authored to DVD-Video at 640kbps in clear violation of the specs (Pink Floyd's "Pulse" Live album did this) which is complete stupidity. WHy not simply add a half-bitrate DTS stream? Ask Sony, I guess. But again I digress, and wander OT. Sorry about that.
If your client is providing you with 320kbps streams for stereo, this ought to work just fine. But your original post did heavily imply that Encore was doing the encoding for these streams, hence my comment.
>(2.0 track is 320kbps (Encore encodes the 2.0)
As far as Replication houses rejecting masters goes, this is a whole can of small, pink wriggly things all of it's own, and the blunt fact of the matter is that replication houses
i almost never bother to correctly verify the actual streams in your Video_TS.
At best they will run the DDP files through the Eclipse system, and as all the various warnings are user-settable, what they will flag as an error is very much a case of pot luck. About the only way I know of to absolutely validate & verify your streams is to use the MEI DVD-Video Verification tool in combination with the full DVD-Video specifications, or else use a service such as Trai Forresters to get the job done for you. See http://www.dvdverification.com/public/157.cfm for a complete description of these services as well as a very good explanation as to why this is a good idea.
To summarize, Replication companies do not verify your data. They do NOT, repeat NOT check your data for DVD-Video spec compliance. What they will do is run the
i image
through Eclipse, and may or may not act on the warnings & errors it flags. Odds are very high that unless you are using
A - A very reputable replicator, or
B - One that actually gives a damn (rare these days) then whatever happens - unless the eclipse warnings are flagged as FAIL errors, with the ABORT recommendation (and even this is user definable too) then they will most likely still press whatever you send them, and when it all goes pear shaped they will blame you. An image that passes Eclipse does not necessarily comply to the DVD-Video specifications. That is your job as the authoring house.
Having said that, I would be extremely wary of attaching total reliance to any freeware utility in this matter. However, it would seem that there is a doubt over the true maximum bitrate of your stream, and this may well cause playability problems in some players. Not all, as not all DVD players are created equally, sad to say.
I strongly suspect I have just scared the willies out of you now. That was not my intention, and I apologise if this is what has happened.
Let me try to go into some more details here without repeating verbatim what you can go & read for yourself on Trai's website about this. In the earlier days of this game, replicators were a lot fewer in number, Eastern Europe, Asia & China had not got into the game, margins had not been cut to the bone to stay open & operational, and discs got checked & verified a hell of a lot better than is common these days. Problem is it can get bloody expensive. Deadlines loom, clients start getting twitchy, but the accountants keep wanting more & more for less & less and threaten to move contracts around of they do not get "discounted services". So companies start to cut corners.
But again I digress and wander, and again I apologise. Most of this information in this post will be irrelevant to your project, and will not be anything to worry about.
The point of making this post was not to scare the bejesus out of you, but to point out that you simply cannot rely on the replication house to tell you if your streams are out of spec - they simply will not do that any more than they will flag VOB errors in the navigation code either. This is just not something they ever check.
The problem is that all authoring packages - there are no exceptions here at all and some are worse than others (with Apple's DVDSP being one of the worst offenders) - interpret the DVD specifications in their own manner, and can easily produce non compliant streams. If you are worried about a video file, you need to either triple check that file, or else re-encode it.
Let's look at the numbers on your files again, just to be certain I have got all this correct.
2 Audio Streams. 320 & 448 = 768kbps.
1 Video Stream, VBR, Peak to 8700kbps.
If this is correct, IMHO you are pushing your luck somewhat. That adds to 9468kbps, leaving zip overshoot room for any unexpected spikes in the bitstream caused by sudden shifts in the material - and in 2-pass VBR this
i will
happen more often than you think.
What are you writing your master to, please? (I really want to hear "DLT Tape" and not "DVD-R Disc" here as well)
How long is the video stream?
What is the average bitrate of your video file as set? If the target is as stated, 7500, I would forget all about using VBR here and encode to a fixed bitrate (assuming you have the space to do this, I would set the CBR to around 7,000 maybe 7500 and master to DLT, as at that bitrate a written disc may choke in players anyway, regardless of stream spec compliance.
One final thing - and this comes from personal experience too - is to use the best MPEG-2 encoder you can get your hands on. MainConcept is okay for a lot of things, but it struggles in certain circumstances.
It
i can
also give peaks that are well outside what you
i think
you have set as your upper limit.
You mention that you used MPEG Validator. BUT - and this is a biggie - if one says PASS and the other (Bitrate Viewer, which is no longer being developed either, so is way out of date) says FAIL, I would look for confirmation somewhere down the line. MPEG Validator will - and I am sure you already know this - only validate the streams in PAL, not NTSC. There are varying reports as to how accurate it really is.