Is skipping/jumping due to bitrate spikes?

[absinthe]

A DVD I've (re)authored from my latest encode jumps and skips at certain points, so I'm wondering: What causes this phenomenon? Is it a result of bitrate spikes?

I can't imagine what else would cause it, but I've been using the same settings in CCE for quite some time so maybe I need some tweaking.

A few details: The movie is an interlaced DVD rip (movie was apparently shot on video). I do 4-pass encodes, and traditionally I've always checked "close all GOPs." I've noticed that many tutorials leave this unchecked. I took a look at the first of the movie's VOBs (from the original DVD) in Bitrate Viewer, and only the first GOP is closed; the rest are not. Exactly what purpose does closing a GOP serve?

Also, the first VOB's peak bitrate is 5538, average 4749. My encoded m2v file has a peak of 8074 and an average of 3988. I've got three AC3 audio files and one subtitle file on my reauthored DVD (audio bitrates 192, 192, and 448). In CCE, I generally set max to 9500, minimum 2000, and average as calculated. All my other settings are pretty much the standard recommended: M=3, N/M=4; SEQ header every 1 GOP; add SEQ end code checked, and that's about all under the Video tab.

I also used the bitrate viewer tool in DVD Lab Pro (trial), and it never went over what was indicated as max.

I'm giving it another shot, and I'm going to try to tweak the bitrate a bit ala one of the handy-dandy doom9 tutorials. I'm also going to notcheck "close all GOPs."

Anyone have a guess at what could cause this? I don't think it's the DVD Lab Pro beta. I've authored several discs in the couple of weeks I've used the trial, with multiple audio tracks and subs, and they've come out perfect. And I also burn my discs with Nero in UDF/ISO mode.

thx,

-abs

[absinthe]

An update:

I've educated myself on using the bitrate distribution tool in CCE, and there are most definitely grey areas (i.e. not green) that perfectly correspond to the "skipping" parts of the movie. However, no matter how much I tweak the local bitrate settings, I can't seem to fix everything. Fantastic tool ... but what a pain! :)

I'm wondering something: DVD2AVI reported the MPG from the DVD as interlaced, but in watching it I can discern a definite 2:3 pattern. I tried the SmartDecimate filter (a favorite of mine), and it seems to have worked nicely.

I wonder if I used this, and thereby reduced the number of frames, if that would increase the overall "Q" factor in these tough spots.

I'm tempted to just throw in the towel and make a darned XviD of it (just bought a Philips DVP-642), but I hate to quit after spending all that time designing menus, etc. . . .

-abs

[br408408]

"Exactly what purpose does closing a GOP serve?"
For Editing.

What other software are you using?

[fccHandler]

From what I've read, the combined bitrate of all streams in a VOB should not exceed 10080 kbps. But if you add your audio bitrates to your maximum video rate of 9500, it becomes 9500 + 448 + 192 + 192 = 10332. Whether it really ever hits that rate is questionable (your bitrate viewer doesn't seem to think so). But just to allow some margin of safety I would try lowering your maximum video bitrate to 9000 when you have that much audio.

Also try checking the "DVD Video Compliance" box if you haven't done so already. IIRC the CCE manual says some stuff about how this box affects VBV and frame sizes (in bytes), and I always use it even though some guides claim it's unnecessary.

[br408408]

Sorry, I posted at the same time you posted your 2nd post. DVD2AVI will sometimes drop frames. Try DGIndex...it might help

[absinthe]

Originally posted by br408408
Sorry, I posted at the same time you posted your 2nd post. DVD2AVI will sometimes drop frames. Try DGIndex...it might help

Yep, already using DVD2AVIdg.

-abs

[video_magic]

Originally posted by br408408
"Exactly what purpose does closing a GOP serve?"
For Editing.

What other software are you using?

Hmm, I have not find a definitive answer for this.
I believe there is a problem with knowing the DVD specs, because one has to pay to see them and sign a non-disclosure agreement. But I have read a few places that closed GOPs are a requirement for some standalone DVD players to play the disc properly. I wish I knew whether 'closed GOPS' was actually a requirement to be spec-compliant, does anybody here know??? :(

Additionally, the bitrate average (of all video and audio and other data added) outputted from the buffer should not exceed the 9.8mb/s I think, otherwise you could get playback issues, once again I wish I knew for certain.

Maybe closing or not closing GOPSs affects playback/rate control because of frames referencing frames in other GOPs. I am hinting that maybe a GOP is somehow measured as a unit for data rate, and this could be confused if it's not closed to be a proper independant unit but references frames outside its' self.

By the way, I am also now confused whether a Key-Frame and an I-Frame are the same thing, because of something I read. I know marking a frame as a Key-Frames is used for seeking to, but I had not previously thought this was meant to affect quality. I.e I thought if there was a sequence IPPBBPP and I specified a Keyframe every 3 frames, then the I frame, first B and last P frame in that little sequence would get marked to be seekable -nothing to do with quality of the frame. However, here is a quote from Mencoder documentation

"keyint=<0−300>

maximum interval between keyframes in frames (default: 250 or one keyframe every ten seconds in a 25fps movie). Keyframes are needed for seeking as seeking is only possible to a keyframe, but keyframes need more space than other frames, so larger numbers here mean slightly smaller files but less precise seeking. 0 is equivalent to 1, which makes every frame a keyframe. Values >300 are not recommended as the quality might be bad depending upon decoder, encoder and luck. For a strict MPEG-1/2/4 compliance this would have to be <=132.

and then also

sc_threshold=<-1000000−1000000>

Threshold for scene change detection. A keyframe is inserted by libavcodec when it detects a scene change. You can specify the sensitivity of the detection with this option. -1000000 means there is a scene change detected at every frame, 1000000 means no scene changes are detected (default: 0).

(edit: in this scene change detection I had read elsewhere that an I-Frame, not a Keyframe, would be inserted so I wondered if the two kinds of frame are somehow interchangeable)

So, I am asking if a keyframe has to be an I-Frame, but however an I-Frame does not have to be a Keyframe.

I welcome any corrections, (preferably simply worded :) ) explanantions, and advice. I am trying to make sense of what I have read by searching.

[video_magic]

Little update, although this only confuses me more. I found a page about Mpeg-2 and here is a quote, but it seems to be saying that GOPs are only useful if you are using B-Frames?, anyway here's a quote from this page:
http://www.mpeg2.de/doc/mpegfaq/mpeg2.htm

Group of Pictures:
The concept of the Group of Pictures layer does not exist in MPEG-2.
It is an optional header useful only for establishing a SMPTE time code or for indicating that certain B pictures at the beginning of an edited sequence comprise a broken_link. This occurs when the current B picture requires prediction from a forward reference frame (previous in time to the current picture) has been removed from the bitstream by an editing process. In MPEG-1, the Group of Pictures header is mandatory, and must follow a sequence header.

[Boulder]

I-frame is also known as a keyframe. Mencoder can also output MPEG4 so that's basically why there's an option to do extra long GOPs. If you have tried seeking in MPEG4 and MPEG2 files, you know which one works a lot faster. MPEG4 usually has avg GOP length of over 100 frames whereas MPEG2 standard says max 18 frames per GOP for NTSC, 15 for PAL.

[Arky]

If you're authoring for DVDR distribution, then you'll need to slash at least a couple of mbps off the Spec-defined max combined bitrate because DVDR is lower reflectivity than replicated/pressed silver discs and many players won't cope with such a high rate of data retrieval.


Arky ;o)

[Boulder]

Depends on the media and probably the standalone too. I've lots of encodes with max bitrate at 8500kbps -- which is also being used at scenes with high motion, and no skipping.

[Arky]

Yes, but your player may well use a cheap ATAPI drive - many higher quality players use purpose-made DVD drives, with integrated DSI circuitry. These drives are superior in many respects (for the specific application of a standalone DVD player), but one shortcoming is that they lack the ability ATAPI drives have to produce phenomenal rotational velocities, for example. A common characteristic of purpose-made DVD drives is that their integrated DSI circuitry results in lower latency when dealing with complex navigational structure in a VOB (e.g. interleaving, where Multi-Angle or Seamless Multi-Story is concerned). Because ATAPI drives lack this close integration, their implementations in standalone players frequently involve considerably larger data buffers (irrespective of the meagre buffer requirements of the DVD Spec itself) to cope with the higher latency in dealing with complex playback. This larger buffer has an additional benefit in that any circumstance where the disc proves difficult to retrieve data from, the ATAPI drive is capable of upping the rotational velocity very considerably, and the larger buffer provided in the player is able to capitalise on this capability to better cope, in sum, with such a poor reflectivity disc (incidentally, this also applies to playback of DVDs on a PC/Mac, for the same reasons).

This is just one reason why some players cope differently to others when they are asked to retrieve high bitrates from low reflectivity formats (e.g. DVDR), even though, on paper, there should not be an issue, since even a purpose-made DVD drive should theoretically have ample capability to retrieve data from even low-reflectivity media, and the player should, at least as far as the DVD Spec itself is concerned, have enough buffer memory. There are, of course, other contributory factors to the ability of a player to retireve data from low-reflectivity media - I am simply pointing out that players differ very significantly from each other 'under the hood', so the fact that you may have successfully played back high bitrate DVDRs on one machine does not mean this would be the case on all standalone players.


Arky ;o)

[Boulder]

I'm quite sure my player uses a specific drive instead of PC technique, being an old (something like 4 years I guess) Pioneer DV-343. The funny thing is that it's very strict with SVCD bitrates but not with DVDs.

Yeah, I've seen some people actually change the DVD drive to a more silent one, the cheap ones are pretty much all PC hardware.

[Arky]

Originally posted by Boulder
I'm quite sure my player uses a specific drive instead of PC technique, being an old (something like 4 years I guess) Pioneer DV-343. The funny thing is that it's very strict with SVCD bitrates but not with DVDs.

Yeah, I've seen some people actually change the DVD drive to a more silent one, the cheap ones are pretty much all PC hardware.


Yes, that is interesting. SVCD, being written on CDR, does place relatively greater emphasis on rotational velocity, owing to the less-densely-packed nature of the data on a CDR, versus a DVDR, so your player's fussiness with SVCDs could be interpreted a number of ways. If we assume that it is a purpose-designed drive, rather than an ATAPI one, then this might suggest an inability of the drive to cope with the higher R-V required to read excessive bitrate from a CDR. Considering this possibility in light of the player's ability to cope with high bitrate from DVDR, this might be attributable to an over-specified buffer, a generally-well specified laser and servo assembly, and a very-well-implemented DSI stage. Pioneer are certainly known for overspecifying certain elements of their players (a positive thing), so this might be the case in your circumstance. These two hypothetical behaviours, despite common hardware (whatever that hardware may, in fact, prove to be) are not necessarily at odds with each other - the players firmware itself plays a very significant role and differences in performance, on certain factors, across differing media types are very common, according to the specific strategy and resource 'tuning' implemented by the designer of your player.


I'm glad you can happily play back high bitrate DVDR on your player - you are fortunate. I made my earlier remarks purely from the standpoint that many people author for duplication, so it is important that they are aware of the potential pitfalls of authoring with high bitrate for DVDR distribution, even if the DVDR test disc works correctly on their own test player. Verification (http://www.intellikeylabs.com/pages/services_dvdtest_page.html) is a huge industry in itself, and with good reason! :(


All the best,


Arky ;o)

[hank315]

Hmm, I have not find a definitive answer for this.
I believe there is a problem with knowing the DVD specs, because one has to pay to see them and sign a non-disclosure agreement. But I have read a few places that closed GOPs are a requirement for some standalone DVD players to play the disc properly. I wish I knew whether 'closed GOPS' was actually a requirement to be spec-compliant, does anybody here know???The next link may give the answer, it's the MPEG ISO standard, Columbia University has it online:

http://www.ee.columbia.edu/~eleft/e6880-Spring98/docs/is138182.pdf

Also this document is very useful:

http://www.mpeg.org/MPEG/DVD/Book_B/Video.html

[dragongodz]

actually the only time i read that closed GOP is a must is for mutli-angle dvds. i have not seen it mentioned as being required for dvd for anything else.

it is of course needed for accurate editing of streams aswell but thats slightly OT. :)

[video_magic]

Thanks guys :)

Thanks also for the links Hank.
That PDF looks pretty heavy reading, but I'll take a stab at getting my brain round some of it! :scared: