Are DVD max bitrate guidelines outdated?

[zambelli]

Many documents and web pages about DVD authoring recommend that the max bitrate of the MPEG-2 video stream should be set at no more than 8 Mbps, some even going as low as 7 Mbps. The typical explanation is that "many DVD players can't handle high bitrates".

Now, the DVD spec clearly states that the combined max bitrate is 9.8 Mbps, so it would seem that it'd be perfectly fine for the video bitrate to go up to 9.5 Mbps if the audio stream was only 384 kbps, for example. So why the 7-8 Mbps max bitrate recommendation? Is it just an outdated recommendation from the days when most DVD-ROMs were 1x devices? I don't understand why 9 Mbps would be a problem for any modern DVD reader or decoder.

[setarip_old]

I don't understand why 9 Mbps would be a problem for any modern DVD reader or decoder.I rather doubt that this would be a problem nowadays.

On the other side of the coin, most commercial DVDs that I've watched over the years on a SONY player, with bitrate being displayed, have average bitrates around 5Mbps. Of course, there are occasional significant surges (High action, explosions, etc.)...

[LadyLiete]

Bit rate control is best... keep it 5, except in heavy action jump up to 7 or 8.

[dragongodz]

the DVD spec clearly states that the combined max bitrate is 9.8 Mbps, so it would seem that it'd be perfectly fine for the video bitrate to go up to 9.5 Mbps if the audio stream was only 384 kbps, for example.
actually its 9.8 for video only but once you start adding other streams the combined(muxxed) 10.08 value is really the more prevelant. this includes not only the streams bitrates but muxxing overheads etc.

I don't understand why 9 Mbps would be a problem for any modern DVD reader or decoder.
for the vast majority it wouldnt. however there are several things you need to consider.
1. if being done commercially you do not want returns, even from people with older or dieing or plain flakey hardware. setting a lower maximum can help avoid this.
2. if writing a general guide again you want it to be as compatible with everything out there as possible. again reccomending a slightly lower max helps there and avoids emails complaining "your values didnt play properly" etc.
3. NTSC pulldowned material. the bitrate in this case seems to be taken from displayed frames and not encoded frames. so while 23.976fps as 23.976fps may be under the limit once you apply pulldown to make it 29.97fps the max bitrate increases to the bitrate of the 29.97 frames shown. if you follow. so in that case you must set the max bitrate lower to allow for that increase.

[mpucoder]

3. NTSC pulldowned material. the bitrate in this case seems to be taken from displayed frames and not encoded frames. so while 23.976fps as 23.976fps may be under the limit once you apply pulldown to make it 29.97fps the max bitrate increases to the bitrate of the 29.97 frames shown. if you follow. so in that case you must set the max bitrate lower to allow for that increase.not true, the bitrate maximum is for the encoded stream, and pulldown, in fact, allows for higher peak bitrates due to an even greater amount of time for 2 of the 4 pictures to refill the buffer.

@zambelli: while there may have been some 1x DVD-Rom drives (I never saw one, my first drive was a 2x) no standalone ever used such a slow device. Standalone players must be able to skip rapidly for trick play and multi-angle. Speaking of which, multi-angle does have a lower max bitrate of 8.0Mbps for everything (all streams and mux overhead)

[dvdboy]

I always encode around 7-8Mbps, but mainly because most of my work goes to DVD-R which I believe may cause issues at the maximum bitrates (unlike pressed discs which the spec was based upon)

[SeeMoreDigital]

All my "Superbit" DVD's seem to work okay, even in my cheapo DVD players!

And from what I understand Superbit DVD's are still being pressed: -

http://www.amazon.com/s/ref=nb_ss_d/105-3253793-5362065?url=search-alias%3Ddvd&field-keywords=Superbit&Go.x=7&Go.y=10


Cheers

[dragongodz]

not true
well to be more exact i was refering to past posts, both here and elsewhere, about pulldowned material erroring in authoring programs. while encoding with a lower max bitrate was accepted.
ok these are older versions and maybe the newer versions of scenarist and maestro(which i remember were 2 mentioned) do not have this problem any more. i do not own either so can not verify.

really though if you are just doing them for yourself you can encode how you want and if you run in to any issues fix it. however, as i also said, for a commercial company or general guide that is something to be avoided as much as possible. advising a general lower max bitrate is such a way to avoid feedback and returns etc.

[Sir Didymus]

...i was refering to past posts, both here and elsewhere, about pulldowned material erroring in authoring programs...


Quoted, quoted!

Bravo dragon. That's IMHO the key factor. In past years people was using the most various (and unstable) authoring and muxing engines. Many of those were simply poor or (in some cases) unbelievably unaware of some central aspects of the authoring tricks and practices - an example is a proper implementation of a VBV model inside the authoring application.

I would add MuxMan to the two example you made of excellent authoring applications available today. Release 0.15P is also free.

Cheers,
SD

[Sir Didymus]

I always encode around 7-8Mbps, but mainly because most of my work goes to DVD-R which I believe may cause issues at the maximum bitrates (unlike pressed discs which the spec was based upon)

LOL. Are you saying that your DVD-R are produced with an authoring application that isn't based on the specs ?

I suggest you to trow immediately this application in the trash!

If you adopt a proper encoder and a proper authoring applications, yours DVD-R are going to become as compliant as the pressed ones!

[Sir Didymus]

All my "Superbit" DVD's seem to work okay, even in my cheapo DVD players!


And this shows clearly the relevance and the (sometimes neglected) high value of the standards: if my cheap player have the DVD-VIDEO logo, I am sure that using a compliant encoder and a compliant authoring tool, I can produce a title with the same bitrate as a printed DVD, and I will have no problems in the playback...

[dvdboy]

LOL. Are you saying that your DVD-R are produced with an authoring application that isn't based on the specs ?

I suggest you to trow immediately this application in the trash!

If you adopt a proper encoder and a proper authoring applications, yours DVD-R are going to become as compliant as the pressed ones!

Hey Sunshine, I've been in this 'business' long enough to see client DVD-Rs misbehave because they've maxed out a DVD at 9.8Mbps (DVDSP 8Mbps CBR + PCM), and they wonder why it's stuttering.

When I'm producing 3500 corporate DVDs on a weekend turnaround, I'm not going to let my Spruce Maestro / MPX3000 encoding go near 9Mbps, because I have seen issues with recordable media far too often.

DVD-Rs are far most susceptible to read issues than pressed discs.

[dragongodz]

a proper encoder and a proper authoring applications
very good point aswell. not only have some authoring programs had problems in the past but of course encoders can also aswell. things like exceeding max bitrate stated springs to mind. i have been able to make tmpgenc, HC, Mainconcept and QuEnc all exceed maximum with certain footage. also its been said, for example in the DVD-RB section some time ago, that CCE will even some times exceed max bitrate.
so that would be another reason to reccomend, especially in a basic or newbies guide, a lower max bitrate.

I'm not going to let my Spruce Maestro / MPX3000 encoding go near 9Mbps, because I have seen issues with recordable media far too often.
hahaha well i can only go by what Mug Funky has said about the MPX3000 and its crappy rate control. i wouldnt let it go too near 9Mbps either but that wouldnt be because of dvd-r speeds. :)
i did however already state for commercial(mass) production its understandable to use lower to avoid even the smallest chance of problems.

[dvdboy]

Hi Dragongodz,

You're right about mass produced titles (My comments were soley targeted at Sir Didymus). I remember reading that ANTZ Region 1 went through at a very high bitrate, and I've never heard of any major playback problems.

If you've got a client shouting at your about returns, it's not going to matter how much you wave the DVD Spec at them!

If you're doing discs for yourself / a small group then obviously you can push the limit of the spec, as you've only got to work on a limited number of machines.

Fair point about the MPX Card, it may not be the best on the market, but I've never had any major problems with it on the 3000+ jobs I've put through it.

[mpucoder]

I've heard about problems with DVD-R too, that they can be more difficult to read than pressed DVDs, therefore requiring re-reads and/or error correction. Isn't there a big thread at this site about the soft error rates of various media?
Personally I've never had a problem, but I test on DVD+RW, how do you guys rate that media?

[mpucoder]

I remember the days of the mplex based authoring programs. what a mess, the program was written in PAL land by someone with no knowledge of pulldown. As a result NTSC material stuttered like crazy, if it muxed at all, as the program tried to add 5 extra pictures per second.

There are two models for buffer management. VBV is the model used by encoders, and for DVD purposes has no time constraint, just a maximum number of bits per picture. The buffer model that authoring programs must use (again, mplex did not) is called P-STD (Program Stream Target Decoder). It is described in the mpeg standard, but certain DVD specific factors such as buffer sizes and critical delivery times, are obviously missing from that standard. The reason the VBV model has no time constraint is that is a problem for the multiplexor to solve.

The multiplex rate for SD-DVD non-interleaved is 10.08Mbps, so it's not possible to stuff more data in than that. HD-DVD multiplexes at exactly 3 times the SD rate (30.24Mbps) and there are good reports of HD_DVD on recordable (SD) disks.

[dragongodz]

soft error rates of various media
another factor to the mix. indeed different media can give different results. for example a friend of mine some time ago started putting his family home videos on to dvd. the video camera he bought had cables and software to do it infact. however initially he had playback issues etc. when i saw the cheapo nasty dvd-rs he was using the first thing i told him was to buy better ones. amazingly the problems disappeared once he did.
now even with good quality media you can expect the odd one to be iffy but it should be more rare than common. so if you were planning on long term retention, less problems and issues or mass reproduction you are best making sure you get the good brands etc. unfortunalty mr average joe will tend to look at price first.

so for a quick short list we have - encoding software, authoring software, media, playback hardware. if any of these perform less than they should it opens the potential for someone to complain it doesnt work. that is out of all the people that read the reccomendations or basic guides i mean.

[Mug Funky]

FWIW our max BR with the MPX3000 is 8500 here - at least that's the rate arrived upon by us.

and yes, CCE will spike if 1 pass was used (for CBR or OPV). another pass will get rid of most of the spikes, but it seems to still allow some.

pulldown should allow a higher bits-per-frame (which in programs like bitrateview that aren't pulldown aware will read as an actual higher bitrate, as it's reading 29.97 instead of 23.976). you can confirm this with the MPX3000 in IVTC mode, and bitrateview. it goes really high on the film parts.

a comprehensive Max Bitrate guide would be very ugly - it'd have to include the fact that max bitrate is measured differently program-to-program (both encoding and authoring), so what passes one test will fail another, and what plays in one player might well fail in another, even if the bitrates "technically" (whatever that means) were within the max.

spruce measures max bitrate per-GOP (that's what it looks like), and HC and CCE do it this way too (CCE is switchable to another mode actually). the MPX3000 does it in it's own strange way that only makes sense to it and spruce's muxer. quenc and streameye seem to do it per second (though DGZ will be able to say more on that).

[kumi]

spruce measures max bitrate per-GOP (that's what it looks like), and HC and CCE do it this way too (CCE is switchable to another mode actually).
Hi Mug Funky, do you know if the bitrates reported by DVD Rebuilder during prepare phase are also per-GOP (using HC or CCE mode)?

Also, would you happen to know what method DVD Bit Rate Viewer (http://www.visualdomain.net/bitrate.htm) uses? Everytime I compare it's reported average bitrate to the reported "typical" bitrate in DVD Rebuilder, the latter is consistently a higher number.

[dragongodz]

quenc ... seem to do it per second (though DGZ will be able to say more on that).
correct.

max bitrate per-GOP (that's what it looks like), and HC ... do it this way too
hmm well no it shouldnt actually. ok i havent talked to hank315 about this part for a long time but from memory the way HC works is per GOP gets a maximum of the percentage of frames to a second percentage. ie. if the number of frames in a GOP is 70% of the frames required for 1 seconds footage then the max bitrate for that GOP is 70% of the max bitrate set.

as i said though that was ages ago with the earlier versions of HC so the exact functioning of that has probably changed a bit.