DVDSP says "Video Bitrate Too High", but why?
 

I've managed to get my multi-angled tracks' GOPs conformed, etc... now I am getting a "Video Bitrate Too High" message upon building my DVD.

The track (track 3) that the error is occurring for has three video streams, all of which were encoded with 2-pass VBR (average=4.7; max=7.0). According to DVDSP's inspector window, the average Bits/Second for the video stream is:

V1 = 4,998,041 (or 4.76 Mbps)
V2 = 4,994,258 (or 4.76 Mbps)
V3 = 5,001,921 (or 4.77 Mbps)

There is one audio stream, a 16-bit/48k AIFF with an average data rate of 1,536,000 bits/second (or 1.46 Mbps)

According to the DVDSP (4.1.2 is the version I am using, FYI) user manual, on page 70:

Though they are very close, it is my third video stream that is the highest. So V3 + A1 = 4.77 + 1.46 = 6.23 Mbps, leaving me with a 1.77 Mbps margin.

What do you think is going on here?

Remember that when using alt-angles that the total amount of data per track has to be accounted for included the angles. A max bitrate of 7 is too high, try a max of 6 to 6.5 with an average of 5.

Get the Apple Training Series DVDSP4 book; you'll get all the answers and troubleshooting tips you need.

I don't understand how I'm not doing that. It says to calculate using the "highest bit rate video stream + all audio streams + all subtitle streams" which, in my case, is 4.77 + 1.46 = 6.23 Mbps (there are no subtitle streams). Is my math bad? Or am I missing something else?

Don't know why, but...
 

Replacing the AIFF audio streams with their AC3 counterparts solved the issue. However I still don't understand why it was telling me the video bitrate was too high. It's not making sense.

@Robert: that book is certainly on my shopping list!

Actually now it makes sense to me; although we normally consider audio files to be small both in size and data rate compared to video, in the MPEG-2 stream a non-compressed audio codec really takes up a lot of system resources during playback, and when you add multiple streams or angles that's an even heavier load.

DVDSP4 while not 100% able to author every aspect of the DVD spec is built around the DVD/ISO specification which, by preference wants to see a compressed audio codec for proper playback. In reality, your build with AIFF's *might* have played fine on newer players, DVDSP4 knows the limitations of the ISO spec and has built-in limits to prevent a bad build-out.

I've seen in extreme cases where code-monkeys have hacked DVDSP4 to lose the various warnings when bitrates get too high in an attempt to maximize playback but they almost always create playback issues with most set-top DVD players.

In short, DVD audio needs to be 2 things for both a good encode and for proper sound: it needs to be the AC3 (or other compressed format allowed by the DVD spec) and max/peak levels need to be preset in editing at -12db. (You can also use Compressor's built-in audio filters to bring the audio down to the -12db level if all your FCP sequences were edited with audio peaking at maximum.)

Trust me, once you get the book you'll have so many "I didn't know that..." moments you'll wonder how you ever got along without it.

I sooooo need to get that book!

OK, here's another one for you. I have a successful build, burned fine and plays well in both my Samsung and Panasonic DVD players (and fine on both Macs through DVD Player). The only oddity is playback on the Samsung, one of the tracks (the same one, every time) seems to skip a bit as if it was playing back at 2x speed. Not in the Panasonic, not on the Macs... so I am leaning to assume it is just my one DVD player acting up. The section of the disc where the section of data lays is about the middle of the disc, not either inner or outer rim, and there are no scratches/dust/fingerprints/etc.

I don't know what to do to further debug. I've never seen this before. And I am a bit nervous about sending in my DDPs to the replication house if this is going to turn up on other people's players -- I know there's always the possibility, but I want to make sure I CMA (Cover My A**).

It could be that's where your layer-break is? Try using different media for the burn; Ridata and Verbatim seem to be the media of choice.

Also, the Pioneer DVR-112D or newer seems to be the best DVD burner for the Mac (many MacPro's shipped with that as the Apple-branded SuperDrive).

Unless you've got DVDAFterEdit to help you check the proper build of the actual DVD image there's no way to know exactly what is causing the skipping, but try the above and see what happens.

The layer-break happens at the near end of this particular track track, however the issue arises from the beginning of the track. It perplexes me that only the one DVD player exhibits the behaviour, which makes me want to trust that it builds properly -- but it does concern me.

Verbatim DVD+R DL is the media I am using with an external LaCie FW drive (my Mac came with the combo drive... I'm on a dinosaur dual-867 G4 MDD).

I downloaded the demo, however seems to crash when I try to open the VIDEO_TS folder!

One of the processes during DVD burning or file builds is the "muxing" of the MPEG2 video track and the audio track into one file that conforms to the DVD specs. While either track can contribute to a data rate too high for the now ancient DVD disc standard, DVDSP makes the assumption that the video track is the culprit. Since Compressor uses VBR compression it might say 6.2Mbps in the info box for the video track but it could be peaking at higher than that for various images.

When I make demo discs for clients that are very short, I have created some presets at 7.2Mbps but due to the VBR scheme that Compressor uses I must use AC3 audio since AIFFs will require too much data for the DVD standard. I have yet to experiment with higher CBR compression in the new version of Compressor.

As I say, although the DVD spec allows for AIFF's in playback the ISO standard for a DVD image requires compressed audio especially when encoding for 5.1 or greater.

It's important to know, that the only reason AIFF/CDFF files can playback from an MPEG-2 stream is because DVD players are designed to also act as stand-alone audio playback devices - a built-in CD player if you will, hence the actual DVD ISO image will technically support having an AIFF in the track. However, DVD players were never optimized to have uncompressed audio *alongside* a video track which is why the AC3/Dolby compression format was created almost 20 years ago.

This same scheme applies to the Blu-Ray spec; while bitrates are significantly highly for BR both in audio and video compressed audio is still the authoring standard, period.

That's interesting information. AIFFs encode much faster than AC3 which is why I like to use them when possible (not much anymore).

I wonder why Apple chose to not have a 60 minute preset in Compressor 3. Standalone DVD recorders manage to get around 9Mbps video tracks so it would be nice if I could coax something like that from Compressor.

IIRC, though 9.8 Mbps can be achieved, 8 Mbps is the highest usable bitrate due to limitations of some consumer DVD players.

Well, of course. There is really no encoding going on if your source is AIFF and you store PCM audio on the DVD, because the two are really one and the same (except for sample rate and bits/sample, which are easy to convert. When you want to store AC3 on the disc, however, there is real compression (*) going on in the AIFF-to-AC3 encoding, which requires the use of complicated (and time consuming) algorithms.

- Martin


(*) "compression" as in "reducing the size of the data", not "reducing the dynamic range of the audio"

My old Panasonic DVD recorders made MPEG2 and AIFF audio files at all bit rate setting. I know this since I regularly demux the DVDs for other projects. My newer JVC DVD recorders record in AC3 for all settings. These recorders make MPEG2 video files that have higher bit rates then I can get out of Compressor. Anyone have a formula for Compressor to get the highest video file bit rate possible?

Without knowing the model number you're most likely referring to the combo models that have DVD-VHS-MiniDV built in? Those recorders are designed to capture and record *tracks* not create a fully authored disc with menus, subtitle functions, alt-angles etc.

Important too, is that since those devices burn tracks only bitrates can be higher all around.

Compressor nor any other software-based encoder can replicate what a hardware encoder can do, which is what you've got in those combo machines. In fact, if it were possible (and it's not, so don't worry about trying to figure out how) to take these hardware encoders and grab the MPEG-2 file it creates and then import that into DVDSP4 you'd have a superior looking output.

That's why anything encoded with Compressor, BitVice and other software-based engines will never look as good as what the big Hollywood studios output does - because the studios are *always* using a hardware encoder to create their MPEG-2 and AAC/AC3/Dolby compliant files, period.

If you really want to maximize your encodes, get an XP-based machine and purchase one of the few hardware MPEG-2 encoders and use those files to import into DVDSP4. They're not cheap but currently there's nothing that compares to a good hardware-based encode.

On subsequent builds/burns I have had problems getting the track to play at all on my Panasonic DVD player! The Samsung will play, but seems to stutter ahead a few frames (as if shuttling at 2x), but the Panasonic won't even play the track -- it jumps to the next one instead.

I don't know what to glean from this. I don't have the knowledge to know what cause, or at least possible causes, might be. Of what is this symptomatic? Data rate too high? (I most recently tried a 2-pass VBR, average=4.4, max=6.8) Could it be GOP issues?

The three video streams for this track is 24:12 each, and on my dinosaur system takes me >5 hours to do a 2-pass VBR encoding, which is a bit demoralizing to go through such debugging.

Actually that's essentially what I do by recording with these low-cost machines and demuxing the files for import into DVDSP. Works great, costs the price of a DVD blank and I don't have to spend time and money on a dedicated computer with a hardware board in it.

Most DVDs that people create lack multiple angles, subtitles, alternate audio tracks and slide shows. And many DVDs have less then 1 hour of video material on it. A small request of the Apple programmers who may be reading this, please provide us with some optimized presets for the short form project.

Just to make sure I am giving as much information about my problem as I can:

I was previously experiencing issues with incompatible GOP structures on the very same track. The issue was some variances on where cuts were made between the three angles (front, side, back).

The way this issue was resolved was by outputting the entire sequence for each version/angle (one for the front, side, and back) as FCP movies (without any markers), then brought those FCP movies back into FCP and placed them in new timelines with identical chapter markers. It is these new "workaround sequences" that were then put through Compressor and into DVDSP for the DVD.

Following me so far? Just in case, the workflow went like this:

1. Drills_front, Drills_side and Drills_back sequences are exported as FCP movies, retaining timeline settings, no chapter markers. Filenames are Drills_front_4import, Drills_side_4import and Drills_back_4import respectively.

2. The FCP movies are imported into FCP and each placed into new sequences Drills_front_workaround, Drills_side_workaround and Drills_back_workaround respectively, which already have identical chapter markers.

3. These new workaround sequences are exported directly to Compressor, where they are processed with a 2-pass VBR (latest test was with 5 Mbps average, 6.4 Mbps max) with the "Include Chapter Markers only" option in the Extras tab checked.

The resulting .m2v and .ac3 files are brought into DVDSP. When I compile the build, it goes off without a hitch. When I open the TS_FOLDER in Apple DVD Player, it plays just fine. However, if I try to open the same TS_FOLDER with DVDAfterEdit -- the application crashes (I tried DVDAE on some other build folders I know to be good, and it opens fine... telling me there is an issue with the suspect build). If I burn the suspect build to DVD, the problem chapter will exhibit different issues as I have already described in the quote above.

This is causing me to be unable to make the delivery -- which is making me look like an incompetent amateur. Perhaps rightly so. The deadline has been blown; but I still need to get it finished, and finished correctly.

I really need help.

Does the DVD play in the computer or are you using a DVD player? It's possible that the total Mbps of your three angles exceeds the ability of the DVD player to handle. I have never attempted multi-angle but I seem to remember that the tracks couldn't exceed 7 or 8 Mbps when added together. The last multi angle DVD I played was an extra on a major studio film and each track (there were four multi-tracks) was very compressed. If each of your tracks average 5Mbps then the total is around 15Mbps, too high.

I understand your theory, but I'm not sure it is true.

For 5 angles or fewer, 8 Mbps is the maximum for a track’s combined bit rate. We calculate the combined bit rate by adding the highest bit rate video stream + all audio streams + all subtitle streams.

According to DVDSP manual (page 70):

If it was a matter of having a bit rate that was too high, DVDSP wouldn't complete the build anyway.

I'll plead technical ignorance on how a DVD player deals with multiple angle tracks.

(edited techno-garble)

I plead ignorance no longer. This link provides some helpful information:

http://books.google.com/books?id=98d...l=en#PPA222,M1

In short, the author of this link states that the DVD can switch between the angles live as long as the tracks are burned on the disk within a certain distance from each other so the laser reader head can successfully switch between them. Bit rate is calculated by adding the highest rate video track with the audio track plus tracks such as subtitles, subpictures, etc. Since you have problems before a disk is even burned, I now suspect that one of your tracks is causing the problem. You'll now have to ferret out which track is doing this. Are the FCP movies you made self-contained or just reference files?

You have my sympathies, this is not a good place to be.

Have you tried contacting the publisher of DVDAfterEdit? I'm not sure why it would be crashing on your machine, unless you have directory issues on the boot drive. (time for disk warrior?)

Based on all the steps you've taken I really feel like there's something simple being missed here but unfortunately without being able to see the entire build of your project there's no way to fully ascertain the issue or offer further help via the web.

I'd highly suggest locating a DVD specialist in your area and have them look over your project. You might also check out the Apple forum pages for additional help since there are tons more people doing DVDSP work there than on this forum alone.

http://discussions.apple.com/forum.jspa?forumID=952

I wish I could offer more help but there's too much info that can't be had by just web-descriptions.

Holy Cow!
 

Last night I stumbled upon a workaround. I was convinced that the issue was not with the bit rate, but something else... perhaps GOP, or some other technical issue I was blind to. Nothing scientific; pure hunch.

The problematic track is 24 minutes and 12 seconds in length. I decided to output my sequences in segments. So, rather than having one 24:12 QT/FCP movie for each angle, I broke it up to four shorter (4-7 minutes each) chunks, then placed them sequentially on the track timeline in DVDSP (having gone through Compressor as 2-pass, 5 Mbps/6.5 Mbps average/max bitrate, etc.)

Worked like a charm. I was able to open the TS_FOLDER with DVDAfterEdit just fine, burned and played in all my players without a single issue.

Happy time! I broke open my reserve bottle of El Tosoro Platinum (best tequila... EVER) and enjoyed a glass.

I found a workaround, but still have no answers as to what was going wrong to begin with. Do the methods of my workaround shed any light onto the issue?

I know it is hard to dissect these technical issues across the net, I appreciate and deeply thank you, Robert and William, for lending your aid.