DVD maximum bitrate & multiplex overhead

[FlimsyFeet]

Regrading the DVD spec, I always thought that:

The maximum allowed bitrate for the elementary video stream is 9.8Mbps/9,800kbps.
The maximum allowed bitrate for video stream + all audio streams + all sub streams + muxing overhead is 10.08Mbps/10,080kbps.

Is this correct?

Why is it then, if I try to use Muxman to author one MPEG-2 video encoded at a CBR of 9,800kbps with one AC3 audio stream at 192kbps, no subs, does it fail with an error about excessive bitrate?

I saw a post on another fourm saying that the 9800kbps is actually the max for video + audio. Is that the case, or do I just need to allow for a highr mux overhead?

[bbjay]

Even if you encode your video at 9.8mbps CBR the actual encoded stream is not 9.8mbps all the way through. There is variation throughout the stream sometimes as much as a 1-3mbps (yes even though you are encoding CBR). Google "bitrate viewer" and download that program and run your mpeg and you'll see what I'm talking about.

Now even if you could encode your video at 9.8 no DVD player is going to play it, especially older players. I try to never go above 7.0 for my video if I'm using AC3 audio.

Oh, and 10.08 is the max combined.

Hope that helps.

[FlimsyFeet]

Unfortunately I don't think the figures from BitrateViewer are very reliable (see http://forum.doom9.org/showthread.php?t=140230).
It reports an average bitrate substantially over 10,000kbps, whereas DGIndex on the same file gives the average to be approx 9800 as you would expect.

[bbjay]

Bitrate viewer is the most reliable and accurate viewer. Some only read the header information and thats why you will see the average at 9.8. Trust me I have gone through this whole deal before working with Sonic and Cinema craft. Your best bet is to use a bitbudget and keep you video data rate under 7.0.

Edit: Oh and that should prove to you that the BV is correct since your having problems muxing because of too high of a bitrate. If DGindex was correct you wouldn't have that problem.

[FlimsyFeet]

OK, here's what I found.

If I try muxing the same video with a mono AC3 audio file (96kbps) then Muxman doesn't complain and muliplexes fine. I'll test this out on a DVD player and let you know how it plays.

The sample MPEG2 file I encoded is 186,010 KiB. Lenght is 2m 35.5s. That works out at an averge bitrate of 9799kbps.

VirtualDub with the MPEG2 plugin reports an average bitrate of "9800 Kbps".

Previewing with DGIndex, the bitrate generally fluctuates between 9.6 Mbps and 10.0 Mbps, with occasional peaks up to 10.3 and down to 9.2 Mbps. After previewing the file the average was reported as 9.802 Mbps.

BitRate Viewer shows the bitrate to vary between 11100 to 12200 with an average of 11590. Now I may be reading it wrong, but that does not sound very reliable or accurate.

[bbjay]

Why do need to have the bitrate that high anyhow. Many DVD players WILL have issues with this.

[Arky]

Why do need to have the bitrate that high anyhow. Many DVD players WILL have issues with this.

Yes, especially if you're authoring for DVDR rather than pressed discs.


Arky.

[FlimsyFeet]

A few years ago that would be an issue, but the sample file I burned to a DVD-RW played fine on 3 out of 3 players I tested it on. (Well actually it was one player and 2 recorders.) Modern DVD players are a lot better at reading burned discs than they used to be.

I find that Muxman is good at this. One time I authored a DVD with PCM audio; DVD Lab authored it without complaint but the disc stuttered on scene chnages. Using the same files, Muxman stopped with the excessive bitrate message. So basically, if Muxman can mux it, then it's a good indication that the disc will play fine.

The reasons for using the max bitrate are,
1) the source is an analogue capture, very noisy and difficult to compress, and I want to keep the DVD as close as possible to this (i.e. no filering)
2) the video is short so I may as well use all the space avialable (even using maxium bitrates, there will still be some space left over).

[bbjay]

I guess if this is a personal project and your only going to play it on your 3 players then go nuts. If this is for a client, send it off to a test lab and I promise it will fail miserably.

Yes older players have more issues then newer ones but you have to remember most of the DVD players are older players (DVD player sales peaked in 2003). Not only that but it is up to the DVD player manufacturer to set its limits as long as it provides minimum 100% DVD-Video spec compliance. Also, muxman would not be my first choice for a multiplexer if this was for a client.

Like I said if it is a personal project and you can actually see a difference in quality from 7mbps to 9.8mbps then you have successfully maxed out the DVD spec.

For your reasons of using max bitrate:

1. There are a lot of prefilters you can use to lessen analog noise before encoding (some encoders have them within the program). Which will actually yield a better result.

2. I have done many DVD's containing a couple of 30 second commercials and I would never go over 7.0mbps. Mainly because if this is going out to multiple people you never want a client coming back say "hey that disc you made didn't work!" Plus if you go to replication with this most likely you will get a "data transfer rate too high" error in which it wouldn't even be able to be mass produced.

[manono]

The maximum allowed bitrate for the elementary video stream is 9.8Mbps/9,800kbps.
The maximum allowed bitrate for video stream + all audio streams + all sub streams + muxing overhead is 10.08Mbps/10,080kbps.

Is this correct?
Yes, all that's correct. The reason a CBR encode at the max won't author is because is won't really be a constant 9800. It fluctuates, and you have allowed yourself almost no room for even a little bit of fluctuation. Your DGIndex test alerted you to that fact, and Bitrate Viewer, while inaccurate, does show the fluctuating nature of the bitrate. I'd recommend, if you have to encode as CBR, to lower it to about 9000-9500.
I have done many DVD's containing a couple of 30 second commercials and I would never go over 7.0mbps. Mainly because if this is going out to multiple people you never want a client coming back say "hey that disc you made didn't work!" Plus if you go to replication with this most likely you will get a "data transfer rate too high" error in which it wouldn't even be able to be mass produced.
You can make yours however you like, but most all high quality retail DVDs use max video bitrates of over 7000 (and use VBR encoding). Warner Home Video is among the more conservative with a 7500 max usually. Many go up to and above 9000. To say they can't be mass produced is simply untrue.

[bbjay]

You can make yours however you like, but most all high quality retail DVDs use max video bitrates of over 7000 (and use VBR encoding). Warner Home Video is among the more conservative with a 7500 max usually. Many go up to and above 9000. To say they can't be mass produced is simply untrue.

Not to start an argument, but I have done work for Warner, NewLine, and Starz and none of them ever used bite rates higher than 5-6 most are around 4-5vbr. If it is higher they are using a special encoding (Superbit) and special authoring tools (Toshiba/Panasonic/Sony) which was mainly used by Columbia Tristar.

When I said the thing about not being mass produced I was referring to his 9.8mbps video. I just put it in after I said that "I" never go above 7.0, so that may be confusing.

The fact of the matter is even the audio bitrates fluctuate as well even if they are encoded for CBR. So take your 7.0 CBR video stream and then add a PCM audio track @ 1.5Mbps= 8.5Mbps. Since video encoded CBR can fluctuate around 1-2 mbps you will be over the max once both audio and video stream flux at the same time. The same is true with Dolby 5.1 audio, which mostly all Hollywood DVD's have both 5.1 @448k and a 2.0 track @ 192k there is simply not enough head room to encode your video too high. Trust me these studio's aren't going to replicate 1 million+discs if there is a chance that they won't play in some or even a few players.

[manono]

Hi-
Not to start an argument...
Go ahead, start a discussion. :)
...none of them ever used bite rates higher than 5-6 most are around 4-5vbr.
I'll assume you're talking about average video bitrate and not maximum video bitrate. As I clearly said earlier, I was talking about max video bitrates. If you are also talking about the same, then I might question your claim to be a professional. I can point out hundreds of retail DVDs with max video bitrates of over 6000, although all I'd really need to refute that claim is one. I can also point out some with maxes over 9000 and averages close to that.

Look, this is an important point. I read all the time advice to keep the max bitrate below 5-6-7-8000, take your pick. And one reason given is that some players won't play them. But all are required to be able to play anything conforming to the DVD specs and what possible reason can there be to cripple an encode so it'll play on old, obsolete, and defective players? I'm talking here about encoding for home use where you'll know the ability and limitations of your own player. Sure, the studios have to make some compromises, but nothing to the degree you're implying. They also have the additional advantage of using pressed discs.

The other reason given is that you can't tell the difference between, say, an encode with a max of 6000 and one with a max of 9000. But that's incorrect. One of the primary complaints against the DVD specs is the low max video bitrate allowed. Even using the full allowed 9800 (with AC3 audio of maybe 192), it often isn't enough to prevent artifacts. With a max of 6000 these things begin to show up even earlier. By artifacts I mean mosquito noise, color smearing, and, worst of all, macroblocking during intense and complex scenes. Sure, the professionals have ways to make the video more compressible so these things don't show up. The price paid is a smoother and less detailed video. We as amateur enthusiasts don't always have the same options (although there are some things we can do). So take your 7.0 CBR video stream and then add a PCM audio track @ 1.5Mbps= 8.5Mbps.
Well, as you and I both know, Hollywood DVDs typically don't use PCM WAV audio. And, of course, you have to adjust the max video bitrate depending on the types and bitrates of the included audio, less with multiple AC3 audio tracks, particularly when DD 5.1, and less when using even half-bitrate DTS audio. FlimsyFeet clearly said he was using AC3 audio@192 for his 9800 CBR encode.
Also, muxman would not be my first choice for a multiplexer if this was for a client.
Watch out; them's fighting words around here. Personally, I'm convinced it's as good or better as a pure muxer than Scenarist, Spruce, Maestro, you name it, and I have full confidence that if it authors with Muxman, it's fully compliant. If it's rejected for one reason or another, there's something wrong with one of the streams.
Since video encoded CBR can fluctuate around 1-2 mbps...
I suppose it depends on the encoder used, and I haven't had enough experience using CBR encoding to be able to refute that claim, but in FlimsyFeet's own tests with DGIndex (very reliable), he found variations between 9.6-10.0, or only about .2 in either direction, figures more in line with what I'd expect.

[effort_ac]

This SuperAuthor software can help you solve, but to pay！