Chapter location in final DVD doesn't match DVDA location

johnmeyer wrote on 8/22/2004, 8:15 PM
This is a major bug. I prepare the MPEG-2 and AC-3 files in Vegas. In Vegas, I put markers at the positions where I want chapter stops, and these import into DVDA. The position of these markers in DVDA matches exactly where they should be (i.e., the bug I am describing here is not the problem with the 29.97 vs. 30 fps discrepency).

I preview the DVD in DVDA, and the chapter stops are where they should be.

I then prepare the files, and play the resulting "DVD" on my computer using WinDVD. This is the bug: The chapter stops are sometimes over one second earlier than they should be. I burn a DVD, and the chapter stops are also in the same, wrong places. [Edit: I just went back and measured, and the difference is closer to three seconds on a chapter stop that is only one minute after the beginning of the DVD].

This renders DVDA almost completely useless. Because of this bug, I am constantly getting a few frames from previous scenes when I select a chapter.

I am pretty sure I didn't have this problem in earlier versions. I am running 2.0a, Build 121.

I described a similar bug a few days ago, but I thought at the time that it only had to do with setting in and out points. However, it appears that simple chapter stops have the same problem.

I tried re-loading the chapter stops, but the same problem remains. I have tried manually moving the chapter stops, and eventually I can get rid of the few frames from the previous chapter, but then the new chapter doesn't start where it should.

Sony has recommended many times in these forums that the proper workflow is to prepare the MPEG-2 and AC-3 files in Vegas and then import these into DVDA. Indeed, the whole reason why anyone would ever buy DVDA in the first place is its integration with Vegas. Probably the biggest single element of that integration is the ability to transfer the Vegas markers directly into DVDA chapters.

This feature is broken.

I have been very critical and very frustrated with DVDA for its entire existence, and have posted my criticisms here on many occasions. However, I have stuck with it because of its integration with Vegas. With this problem, the integration is broken, and it may finally be time to switch to Encore, DVD Workshop, or DVD Lab. The other thing that is pushing me this way is the constant frustration with how DVDA completely takes over my system, making it so sluggish, that it is useless for anything else while either a prepare or burn operation is taking place. No other program has this effect.

Comments

gaetan wrote on 8/23/2004, 2:33 AM
I also have problems with markers. Once I inserted a scene selection menu in DVDA 2.0, I want to edit markers but cannot save them. The following window appears:

An error occured while trying to save markers to the media file. This is likely due to the file being read-only or in use by anoither application.

I made sure my *.sfl files are not read-only and i am only running DVDA.
Any info on that issue would be appreciated.
Thanks
Gaetan
PeterWright wrote on 8/23/2004, 4:29 AM
Sorry you're having this trouble John - you've always been a major helper on this forum - I hope a solution looms soon.

I haven;'t experienced this yet - my stuff is mainly PAL - do you think that's a factor?

In the next couple of days I have a 21 chapter PAL project to convert to NTSC - I'll be checking this aspect out carefully ....
johnmeyer wrote on 8/23/2004, 9:15 AM
So far, the only thing I can find that might be unique is that many of the chapter stops that have problems are placed in the middle of a fade to black transition. I can come up with all sorts of theories about GOPs, but I don't know anything about how the software works. The other thing that argues against this is how far off DVDA sets the chapter marks. They are 1-3 seconds from where they should be, and that is 30-90 frames -- a lot more than a GOP.

Haven't heard anything yet from Sony tech support (I submitted bug reports) on either this, or the similar problem I reported early last week.
JackHughs wrote on 8/23/2004, 11:08 AM
I've been doing some reading on this subject and I"ve come across an interesting comment. If the authoring program uses GOP timecode to set chapter points, the error range in finding the chapter points can be +/- .5 seconds.

It appears that this condition is simply something that must be endured unless the project timecode is re-written to frame reference numbers.

Now, having acquired this bit of information, I freely admit my technical ability in this area is not sufficient to act on it. Hopefully, someone who knows will help out.

Jan
johnmeyer wrote on 8/23/2004, 2:42 PM
If the authoring program uses GOP timecode to set chapter points, the error range in finding the chapter points can be +/- .5 seconds.

I'd seen similar information, and I am assuming that it may have something to do with this. However, since DVDA contains code that can both read and create MPEG files, it would seem to me that it would be a trivial task to break apart the GOP that lies on a chapter boundary and then reconstruct into two seperate, shorter GOPs before and after the chapter point.

The whole point of paying hundreds of dollars rather than tens of dollars for an authoring program is to get this kind of refinement.
mikeinjc wrote on 8/23/2004, 7:27 PM
I got the same error message concerning saving the markers. Has Sony addressed this in an earlier thread?
JackHughs wrote on 8/24/2004, 11:18 AM
I think I may be getting closer to cause of my particular scene selection problem. As I understand it, setting in and out points on an mpeg2 title is functionally equivalent to making two cuts. In order to accurately edit an mpeg, every cut must correspond to a GOP header. If not, the resultant edit may display unreferenced P and B frames or, the resultant edit may not display these frames at all.

Assuming each GOP contains 14 frames, each GOP amounts to approximately .5 seconds of display. Therefore, missing the GOP header when setting the in or out point may shift the “actual” in or out point by as much as +/- .5 seconds from the set point.

I assume, but do not know for sure, that DVDA does not incorporate an mpeg stream analyzer. So, DVDA can neither tell us where the GOP headers are located nor can it force our chosen in and out points to “snap” to GOP headers.

Once again, I must disclaim any real ability in this area. My comments are guesses based on incomplete knowledge and information. I’m hoping that someone who really knows this business will tell me whether or not I’m on the right track.
Etibari wrote on 10/22/2004, 9:00 PM
If anyone has heard more on this problem, I'd like to know where. I have been fighting the same problem for the last month. I thought it was me doing something wrong.
JSWTS wrote on 10/23/2004, 3:08 PM
This isn't a 'bug' if you will of DVD-A, but a limitation of mpeg2 video streams and dvd authoring. You can only have a chapter mark at an 'I' frame, which occurs only once per GOP. When you scrub through a timeline or set them in an editing app and export; the authoring app (in this case DVD-A) chooses the closest 'I' frame corresponding to the timecode of the chapter mark. On occassion you can be lucky and the two are actually one in the same (a blind squirrel can find a nut); but the chances are they won't be. The only way to get chapter marks to be frame accurate is to have an mpeg encoder that allows for forced 'I' frame insertion. You generally don't see this feature on encoders under $100 (which are the one you see with Encore, DVD-A and the like). Curiously you can do this however with TMPGEnc. You can't with Canopus' Procoder, unless you use the feature for direct output to disc. CCE Pro (for $1900) will also let you do this.

Unless you are willing to buy an additional encoder, you are going to have to live with this limitation.

Jim
Etibari wrote on 10/23/2004, 6:35 PM
Do you have any more info about how to utilize TMPGEnc for the purpose of forced "I" frame insertion? I do own this program. Does it even interface with DVD-A?
JSWTS wrote on 10/23/2004, 8:41 PM
You could try this link:

http://dvd-hq.info/Compression.html

It has a lot of info regarding the TMPGEnc encoder, and mpeg encoding in general. Look at the heading under 'force picture type'. I haven't used TMPGEnc for a while, but they used to have a trial version that allowed you to use the mpeg2 encoding until the software timed out.

Jim
johnmeyer wrote on 10/24/2004, 12:53 PM
This isn't a 'bug' if you will of DVD-A, but a limitation of mpeg2 video streams and dvd authoring.

It is definitely not a bug, in the usual sense, but my perspective is that it is a limitation of DVD Architect, not a limitation of MPEG-2. This problem is solveable, and DVDA could be improved so that the user didn't have this problem.

Also, it is not something that happens only once in a blue moon. It happens frequently and will affect almost every user's project sooner or later. It is especially noticeable if you use fade-to-black at chapter locations because the MPEG-2 GOP is not likely to start over at the exact point where you insert the chapter.
JSWTS wrote on 10/24/2004, 9:35 PM
The thread started out saying this was a 'major bug'. I was just pointing out to the last poster, and I guess anyone else who's read the thread, it's not a bug. I guess we can all agree on that at this point. There are a lot of things I'd like to have and not pay out the nose for them, but that's just not reality. If the competition is such that features like forced I frame insertion become the standard at this price range, then certainly all the better for everyone. However, you can pay >$20,000 for Scenarist and not get the feature you are asking for. Without it, one has to live with 'close enough' is good enough for chapter placement. Unlike dv footage where a mark can be placed exactly at a frame, the limitation of mpeg encoding is such that chapters can only be placed at an I frame. It is a limitation of the dvd spec and mpeg encoding. Forced 'I' frame insertion is the only way you can get there from here, and you are just going to have to fork out the bucks to get that feature. I suppose DVD-A could offer the feature by adopting a different mpeg encoder than the one they currently have licensed, but in either case, we would ultimately have to pay for it.

Jim
johnmeyer wrote on 10/24/2004, 11:27 PM
suppose DVD-A could offer the feature by adopting a different mpeg encoder than the one they currently have licensed, but in either case, we would ultimately have to pay for it.

That presuposes that it would be difficult to implement. Given that Sony has all the code already developed that can encode and decode MPEG-2, it really wouldn't be that difficult to do. I am sure that the Sony engineers that read this will get ticked off when they read that last sentence ("what the heck does he know"), and I am sure there are some specific implementation issues of which I am not aware.

However, I will continue to press hard for them to do this because when my clients click on a chapter, and they end up seeing frames from the previous chapter, it make ME look bad, and I don't like that.

Finally, I see absolutely no reason why adding this -- or any other new feature -- would be something we would "ultimately have to pay for." Sure, we'll have to pay for any new release, and I certainly expect that. But I think you are implying that this is somehow a fantastically difficult feature that would require excessive development effort which would therefore force up the price. The fact is pricing in software has NOTHING to do with features and EVERYTHING to do with volume and with support. As volume goes up, price goes down, and if a vendor is willing to provide professional support (as opposed to email support or no support at all) then price goes up. The number of features in the program has virtually nothing to do with the price.

Don't believe me? Check out DVD-lab. It has rapdily become the product to beat, and sells for under $200, and that's for the pro version. (Yes, I know, some people claim that it doesn't produce "compatible" disks, but no one has offered any proof, and even those that made these statements about compatibility six months ago now say that the latest release has fixed any small issues). This product has more features than just about anything out there.

JSWTS wrote on 10/25/2004, 7:08 AM
John,

It is a bit of a silly debate, don't you think? (On both our parts?). There are any number of features that DVD-A lacks that are included in Maestro or Scenarist, with the latter several thousands more. The pricing difference for Scenarist isn't for support, you pay extra (beyond the sticker price for the software) for that, and how much depends on the level of service you want. However, if your job depends on it and you're doing Hollywood features, you probably would pay the extra moola for an app that makes rock solid discs.

With that said, Scenarist DOES NOT natively do any encoding--that is EXTRA--like $20,000 extra (or a bit less if you don't buy their high end hardware). Maestro doesn't/didn't do any native encoding--you had to buy that as an extra feature. For that matter, DVD-Lab doesn't do encoding, you actually have to buy the app I noted earlier--TMPGEnc. I may be wrong, but the last I recall Sony/Sonic Foundry had licensed the encoder for Vegas and DVD-A; it wasn't an in house app. If they can get that vendor to do it for free or for minimal cost, then I agree it won't result in any great additional cost, if any. If it's an in-house app, then I guess Sony has to balence all the feature requests and the time/cost of adding them with the what the market will support price wise.

I am not presuposing anything, it may or may not be difficult to implement--it's just currently not an option, and it's not a bug. The only one presuposing anything is perhaps yourself suggesting that "it really wouldn't be that difficult to do." I didn't say it was a 'fantastically difficult feature', it just doesn't exist as an option presently. Maybe it will in the future, but based on DVD-A's update/upgrade path to date, I suspect if we were to see it implemented it would be with the next major upgrade--which we would have to pay for.

If this feature is 'mission critical', then I would do my asset preparation outside of DVD-A/Vegas in terms of mpeg encoding, and I offered a couple of other options and a link. I really was trying to help, not stir up the pot. In fact, I never responded to the initial post, but the last (before all of this back and forth stuff). I gave that poster what I thought was helpful information as to how he/she could work around the issue for now. I didn't say you or anyone else for that matter couldn't ask for the feature from Sony. If your customer base is dissatisfied--get another app or encode differently. There are other apps that allow you to skip between I frames and make a chapter selection that best fits, rather than the one the app thinks best fits for you. There are definitely better encoders. DVD-A's and Vega's encoder does an adequate job, but they definitely are not top of the line (or even close).

Jim

PS I have been using DVD-Lab Pro's version since the second beta release. It's wonderful to play with since it opens up much of what the dvd spec will allow that previously (on the PC side of things) could not be obtained unless you were spending thousands. With that said, I personally wouldn't rely on it for mass distribution, since IMO, it still has bugs to work out. That comes from personal experience, not from a third party source. It, like most apps other than Scenarist, rely on abstraction layer 'insulation' of the user from the complexity of the code involved in authoring. The way the abstraction layer works and is created will have an effect on the final playback of a disc. There are those who make (at least partially) a living off of fixing the errors created by this process (http://www.tfdvd.com/). There is software available, or will be available shortly, for those who absolutely need their discs to run properly. Those who do 'one offs' or short runs don't generally need this level of control, so you won't see many at the level of DVD-A looking for this. For straight forward discs I don't think you would have much of a problem. Those that are more complex and include a fair amount of scripting is another issue. For those types of discs I use Apple's DVD Studio Pro 3 (based on Maestro), and it has it's own set of potential issues (again related to the abstraction layer).