Default to 12-2 not AutoGOP with HC

[laserfan]

In using PRO 1.04 with Hank's Encoder, I'm finding that leaving RB's GOP size to Default does not result in Autogop mode w/HC, instead there is always 12-2 used. Example RB HC.INI:
*BATCH
*PROFILE NORMAL
*INFILE D:\DVD REBUILDER WORK\D2VAVS\V01000300002001.avs
*OUTFILE D:\DVD REBUILDER WORK\D2VAVS\V01000300002001.m2v
*BITRATE 7529
*MAXBITRATE 8529
*WAIT 0
*BIAS 20
*ASPECT 4:3
*GOP 12 2

I think this might be causing problems for my media player. Is there a way to allow Hank's Autogop to work (a workaround), is this a bug to be squashed, or am I missing something?

[spyhawk]

I leave GOP to Default also, and there's no "GOP 12 2" line in HC.INI, using HC016. And in HCBatch window, the GOP is on Auto. Take this line out and rerun it and see what it displays.

[laserfan]

Thanks spyhawk but the HC.INI is a file that is generated by RB-PRO.

I tried it again and still, I get *GOP 12 2 in HC.INI using HC016, and in the HCbatch 0.16 beta window I see "12-2" in the GOP line of "encoder settings".

:confused:

p.s. I searched for 'gop' in the RB-PRO directory (file contents), and the rebuilder.ini file contains a GOP=0, and Eclpro.ini has a AlwaysGOP15=1. That's all I can find, nothing in the WinXP registry. And no .ini file with the HCbatch executable.

[laserfan]

Tried every way I could to stop RB's progress and edit the HC.INI file, to no avail. RB always puts a *GOP 12 2 in the INI file, and HCbatch dutifully encodes with that.

How to fix...

[spyhawk]

p.s. I searched for 'gop' in the RB-PRO directory (file contents), and the rebuilder.ini file contains a GOP=0, and Eclpro.ini has a AlwaysGOP15=1. That's all I can find, nothing in the WinXP registry. And no .ini file with the HCbatch executable.Same settings here. I've no idea why yours behave that way. One work around is to create a BAT file invoking HCbatch with several distinct HC.INI (and GOP line removed) for each segment, but this is a lot of work. So maybe someone knows what's going on.

[jdobbs]

From the change log:

- Modified HC values to preset GOP size to 12 rather
than AUTO for sources that will receive pulldown
flags. This is necessary to ensure GOP time length
parameters stay within DVD specs.

The difference between the two of you is probably the pulldown. Do you want to remain DVD compliant?

[laserfan]

...The difference between the two of you is probably the pulldown. Do you want to remain DVD compliant?I don't get it--what decides whether "the pulldown" is used or not?

I like to try to remain compliant, but why, when Hank says HC's Autogop assures compliancy, do you say it will not be? From Hank's doc:

Running in AUTOGOP mode means the encoder tries to distribute the I, P and B frames in an
optimal manner based on the activity of the frames. Encoding using AUTOGOP will always be
DVD-compliant, max. GOP length is 15.

[jdobbs]

If it is a film and is encoded at 23.976 fps it has to have pulldown flags set so that it can be played back at 29.97. There is no choice or decision -- its a fact of the way it works.

It isn't just HC... it is possible for a VOBU to extend beyond the DVD time limit when values greater than 12 are used with pulldown. There is a thread on the subject here (http://forum.mainconcept.com/viewtopic.php?p=5891#5920). Take a look at your original DVDs and you'll see that FILM sources are set to a GOP of 12 there as well. Same story goes for CCE or any other encoder if you try to go higher than 12 on sources that will be telecined.

Hank has modified the code for v0.16 -- adding additional checks that kinda' go above-and-beyond... and it should be ok to use AUTOGOP in that version. But there is still a significant number of folks using v0.15. That code was put into DVD Rebuilder before v0.16 was even released.

[laserfan]

Thanks jdobbs for that explanation; I can't say that I understand it completely, but I see there is an explanation and not that it's a bug or oversight.

I don't like the effect of 12-2 in my players, so I will see about reverting to 1.00RC6 (i think I still have it here somewhere) if only as a test to confirm what I am seeing (jerky, stutter-y playback of zooms & pans). I finally found that 12-2 was the source of the problem by using DVD2SVCD w/HC16 btw.

For a future release I hope you will give the option to let HC16 run in Autogop.

[hank315]

Some links about this issue:

http://forum.doom9.org/showthread.php?p=714126#post714126

http://forum.mainconcept.com/viewtopic.php?p=5891#5920

[laserfan]

Here's the whole story: I used DVD-RB Pro to re-encode the R1 DVD "Quiz Show" and make a DVD-5 backup that is about 60% of the original file size. I did this not just to backup the movie, but also to convert this 4:3 letterboxed program to 16:9 anamorphic. While the resulting DVD, frame-for-frame, looks MUCH better than the letterboxed program converted at 60% using Shrink, when I play it back using my Pinnacle ShowCenter media players, which use a Sigma Designs EM8551 chip, it plays-back in a very unnatural, staccato fashion, most noticeable in pans and motion.

I made a test clip and did 8 or 9 test encodes using all the options available to me in DVD-RB, including a Sigma-friendly matrix. None of them fixed the playback problem. Note the original DVD does not exhibit this problem on my players at all.

After sleeping on it I decided to try DVD2SVCD, which allowed the use of HC16 also. This test PASSED, the playback had none of the stuttering effect of DVD-RB Pro. The *only* difference of the D2D/HC16 combo was Autogop, vs 12-2 GOPs w/DVD-RB.

So I am inclined to want Autogop w/DVD-RB and not the fixed 12-2 setting, to assure that my encodes can play-back without stutter.

I assume that you have provided these links to suggest to me that I might have a problem of a different sort by not forcing 12-2, but I've not seen any problem with Autogop, nor can I make it fail, while 12-2 looks odd "constantly".

So I am suggesting to have an option to somehow defeat "Always 12-2" in DVD-RB, else I am reluctant to keep using it.

[jdobbs]

In the next version of DVD-RB I will check for the version of HC and will not force 12 frame GOPs on if v0.16 is used. But I think what you are seeing is more likely related to the anamorphic resizing (or something else) than AUTOGOP.

Do you have scene detection on? If so, HC will change the size as needed without exceeding 12...

[laserfan]

Thanks for your response jdobbs. No, I in fact left-out any resizing in all my tests, starting from "RB/encoder defaults". Yes, Scene change was On, since I started with pure defaults and then one-by-one changed the following, every time looking at the dynamic encoder window to verify all settings therein, as it was running passes 1 and 2:

HC16 BEST to NORMAL profile
HC16 NORMAL to FAST
HC16 NORMAL and Jawor's 1CD matrix--HCbatch reports CUSTOM matrix
etc. (VBR bias, One-pass, others...)

I also tried a couple of tests with the Quenc encoder (I don't have CCE or Procoder). Quenc's dialog is confusing because Extreme & Slow was not checked (I used Normal mode), but the boxes adjacent showed 12 and 2. I guess I will look at the VOBs it created but assume it was forcing 12-2 also. Quenc, too, exhibits the jerkiness problem; so it's not unique to HC.

As I indicated above I was getting frustrated because I could not make the jerkiness go away. Only when I used HC with DVD2SVCD and it went away, could I compare HCbatch window settings and the only difference was 12-2 vs true Autogop used w/the D2D product.

I admit & concede that this is likely a Sigma Designs EM8551 issue, and probably exists w/its older chips too incl prolly the Hollywood Plus and Xcard products. It does not appear with my newer ShowCenter 200 which uses the EM8620L. So I don't know if this affects many more than 1 of your customers (me!).

I would be inclined to bury a "force autogop" setting in the ini file or something, for the few of us who are likely to notice this problem, but you are the developer and should do what you want. I for one will look forward to whatever change you make.

:thanks:

[dragongodz]

Only when I used HC with DVD2SVCD and it went away, could I compare HCbatch window settings and the only difference was 12-2 vs true Autogop used w/the D2D product.
did you also try DVD2SVCD with AUTOGOP turned off, so 12-2, aswell to confirm the same problem happened then aswell ?

[laserfan]

did you also try DVD2SVCD with AUTOGOP turned off, so 12-2, aswell to confirm the same problem happened then aswell ?No, I didn't, but I will try this tomorrow! Good idea!

[laserfan]

I couldn't wait; so tonight I set-up DVD2SVCD with an HC.INI at *GOP 12 2 and re-ran the test. The output was smooth as silk! So I've been wrong--it is not a 12-2 vs AUTOGOP problem. There is something else that RB is doing that my player doesn't like.

Nice catch dragongodz. As I sit here though I don't know what to do next. I'm too uneducated (and too tired) to imagine what instruction RB-PRO is giving HC (and Quenc) that might cause the stuttering, that DVD2SVCD is not.

I'm thinking maybe applying DGindex Parse tool to the D2V files and comparing? Argh, neither 1.4.4 nor 1.4.5 will open the RB .d2v files? Guess I will sleep on it and maybe look at it some more tomorrow.

[jdobbs]

...probably exists w/its older chips too incl prolly the Hollywood Plus and Xcard products. It does not appear with my newer ShowCenter 200 which uses the EM8620L It sounds like you hit it here...

[laserfan]

Yah jdobbs but still, why does the output from DVD2SVCD play smoothly, and DVD-RB does not, even when the encoder settings appear to be the same. "Inquiring minds want to know"!

This morning I looked at another RB-Pro disc with my media player (I usually don't do this, I normally play my DVDs using a Sony DVP-NC655P). Since it looks smooth, I'm thinking the problem may originate with the fact that the offending disc, Quiz Show, is a multi-angle. I'm gonna do it over, from scratch...

[jdobbs]

Very probably the player doesn't handle interleaving well -- that happens a lot. Everything isn't DVD-RB's fault you know... did you do the DVD2SVCD encoding in multiple angles? You kinda' forgot to mention that, didn't you?

Inquiring minds want to know how the 12-2 was to blame, then the player was to blame, then DVD-RB was to blame, and now it appears that multiple angles are to blame. :sly:

p.s. (Don't take this answer too seriously ;))

[laserfan]

Very probably the player doesn't handle interleaving well -- that happens a lot. Everything isn't DVD-RB's fault you know... did you do the DVD2SVCD encoding in multiple angles? You kinda' forgot to mention that, didn't you?

Inquiring minds want to know how the 12-2 was to blame, then the player was to blame, then DVD-RB was to blame, and now it appears that multiple angles are to blame. :sly:

p.s. (Don't take this answer too seriously ;))Well, I'm just having a problem, and trying to get help finding it! Sorry if you're put-out by my little problem!

I am your target audience, sir, having migrated to DVD-RB from DVD Shrink. I don't know much about this stuff, and less about how your tool works.

I think my claim is valid. When I apply DVD-RB Pro and DVD2SVCD to the exact same source material, I get two different results, one that is smooth (DVD2SVCD) and one that is not (yours). There has to be a reason, and you are not being a lot of help!

Don't take my growing impatience too seriously!

[laserfan]

...did you do the DVD2SVCD encoding in multiple angles? You kinda' forgot to mention that, didn't you?In fact, what I DID forget to explain was my testing method, that I've made a very small ISO of a piece of the original DVD using DVD Shrink, in Re-author mode ie. Title 1 only, trimmed, with a single soundtrack.

Obviously this does not represent the original DVD structurally, but the original DVD plays smoothly, the DVD Shrink snippet plays smoothly, and DVD2SVCD's converted output plays smoothly, but RB-Pro's output does not.

[jdobbs]

No sense of humour, eh?

The point I'm trying to make is that they aren't the same. If I understand it correctly, DVD2SVCD doesn't do multiple angles... which again means the player is having problems... This isn't surprising.

In multiple angles/interleaving there is more than one view of the video that might occupy the same time space. The two or more portions of the video are interleaved together, so you have a group of sectors for view one, then sectors for view two, view three, and then starting again with view one -- over and over for the length of the shared time space.

The player has to know how to, and have the ability to read portions of the video stream while skipping others. My guess is that the software/hardware in question is having trouble doing that. It could be because of disc speed or fragmentation, or the hardware itself.

The acid test is a standalone player. Does the video output of DVD-RB have a problem there?

[added] Also.... please try the source without any preprocessing using Shrink or any other tool. This throws too many variables into the mix -- and I can't determine where the trouble might be, what might have caused it, or even if there was trouble.

[jdobbs]

Is anyone else out there experiencing anything like this?

[jdobbs]

@laserfan

I'm sorry if I offended you -- I guess sometimes the same choice of wording can say two different things to different people -- I was a little too flippant.

I take all issues raised in the forum seriously and I'm very proud of my record of finding and correcting them as documented here on Doom9. But when a problem is posted without all the associated details is is very difficult to focus and find anything. Here's what I'm talking about.

1. When this thread started all I had to work with was "12 - 2 is causing a problem". You very clearly stated that was the problem. I responded to that.

2. Then I hear that "is likely a Sigma Designs EM8551 issue, ... It does not appear with my newer ShowCenter 200 which uses the EM8620L" So I now guess its something related to that player.

3. Later I find that other encodings don't have the problem so it must be related to a specific disc (Quiz Show) and that multiple angles are involved. I realize that the DVD2SVCD comparison is probably invalidated.

4. Now I find that you have preprocessed the disc through editing and I'm not even sure what we are looking at in terms of a source...

So I guess it's safe to say that I'm also frustrated.

[laserfan]

...When this thread started all I had to work with was "12 - 2 is causing a problem". You very clearly stated that was the problem...Here is what I actually said:

...leaving RB's GOP size to Default does not result in Autogop mode w/HC, instead there is always 12-2 used...I think this might be causing problems for my media player.I was right at least in that RB's "Encoder Default" option is not clear, because 12-2 is not HC's "default" mode, Autogop is. I probably should have dropped this thread around Hank's post #11, as I did "change the subject" after that. My fault.

In your last post, you said:

The player has to know how to, and have the ability to read portions of the video stream while skipping others. My guess is that the software/hardware in question is having trouble doing that. It could be because of disc speed or fragmentation, or the hardware itself.Nope, as I said earlier (in a long post, perhaps you missed this) the media player DOES PLAY THE ORIGINAL INTERLEAVED/MULTIANGLE DVD just fine! It's only the RB conversion it's having trouble with.

I am doing a movie-only backup using 1.04 again (even as I post this) which, because it includes lb-->anamorphic conversion, will take several hours. Maybe there was just "a gremlin" of some sort in the process the very first time-around, and this time it will come-out perfect. If it does not I will likely put the disc away and stop worrying about it!!! For all I know it's just an improperly encoded & mastered disc--maybe I should send it to you as an xmas present! Sorry :p

Enjoy your Sunday jdobbs...for me, win or lose, I have learned an awful lot mucking-about with this stuff the last several days.

[jdobbs]

Hope the rest of your Sunday goes well too... I think I'm giving up for the day. I just finished beating my head against the wall for three days on some new coding... and when I sent it to Rockas for testing it died in a matter of minutes.

Sometimes you eat the bear. Sometimes the bear eats you.

Cheers.

[laserfan]

OK my "re-do from scratch" conversion exhibits the same "stutter-y" problem when played-back via my media player, as I originally thought. I can't tell how the new disc is different from the original structurally (e.g. Cell Type flags are the same, mostly 9s) but it looks fine via my standalone so I am calling it quits--I've deleted all my test files.