DVD-RB 0.98.2 FREE output still interlaced, even when deinterlace is used?

[spiovente_iv]

Hi all. Although I am not a "newbie" when it comes to video encoding, I am a new user of DVD Rebuilder, having only installed it last week... thus I am using the 0.98.2 FREE version, and if the problem I am describing has already been noted and rectified in the Pro incarnation, I apologize for missing it.

Briefly, here is the difficulty: I am backing up my daughter's Sesame Street DVDs, which are encoded as interlaced material. She watches on an analog set at our house, but on computer monitors and LCD screens at each set of grandparents', so I am deinterlacing to produce progressive encodes. (The interlace artifacts don't bother her, but they drive me nuts, and I frequently sit with her while she watches.)

I have tried with multiple approaches, first using the included Deinterlace with DECOMB option, and later using SmoothDeinterlacer.dll in conjunction with Disable "Interlaced", but each method is producing the same effective output, with VOBs in which the Progressive flag is set, but video that is quite visibly - and quite strongly - interlaced. On checking the pre-mux encoded video within the "D2VAVS" output folder, the video is not interlaced, and exhibits the expected progressive frames. (Both VOBs and unmuxed video sources were viewed within DGIndex to verify the presence/absence of interlacing. The VOBs with the Progressive flag set, but still exhibiting interlaced behavior, were then unfolded side-by-side within VirtualDub to verify that the fields were indeed producing separate coherent images.)

Has anyone else encountered this? Is there a reason DVD-RB would be re-interlacing deinterlaced, progressive output? Is there a workaround/fix?

- spiv

[manolito]

This is very weird indeed...

I tried to reproduce your findings, but I couldn't. I used an analog TV capture (Interlaced TFF), selected "Deinterlace with DECOMB" in DVD-RB, and the resulting VOBs are indeed deinterlaced, no combing can be detected any more.

I find it hard to believe that your pre-muxed encoded video segments are deinterlaced as they should, but after the rebuild phase your output VOBs are not. I believe this is not possible. Are you sure you really checked DVD-RBs output folder and not the source folder?

Cheers
manolito

[Video Dude]

You didn't say how big the disc is, but I know many of the Sesame Street DVDs are single layer.

Did you enable "Force Reencoding (for Filters)" in the menu?

[spiovente_iv]

Thanks for the responses... problem solved now, as I'll explain:

First of all, most of the discs are dual-layer. These are the Sesame Street Old School box sets, with a couple of complete episodes and a selection of bonus clips on each disc. 0.98.2(free) doesn't appear to have a "Force Reencoding (for Filters)" option, although on the one disc that was single layer it did ask for confirmation that I intended to reencode.

manolito, I double (and triple) checked everything you suggested, but with the same results each time. I suspect the behavior I was encountering must be related to the fact that each episode on these discs begins BFF, but contains field transitions. I managed to find - after a good bit of digging! - various forum discussions of problems caused by BFF interlacing early in DVD-RBs development.

The simple solution was to build an ISO from the DVD-RB output, and then re-extract the entire contents from the ISO using DVDFab HD Decrypter 5 with PathPlayer enabled. Whatever was slightly off about the VOB and/or IFO structure and causing the odd behavior was corrected by this process (and an actual binary change was verified by comparison of the MD5 hashes for the pre- and post-extraction content; roughly half of the extracted files were slightly altered.)

... so, with hopes that no one else encounters this frustration, I nevertheless document it here, so that if they do, they can spend a little less time banging their head against a wall than I did.

- spiv

[manolito]

First of all congratulations for solving this issue, but still the whole thing sounds like witchcraft to me, so forgive me for digging...
manolito, I double (and triple) checked everything you suggested, but with the same results each time. I suspect the behavior I was encountering must be related to the fact that each episode on these discs begins BFF, but contains field transitions. I managed to find - after a good bit of digging! - various forum discussions of problems caused by BFF interlacing early in DVD-RBs development.
The stuff discussed in these older posts was about field order problems and had nothing to do with deinterlacing. It only occurred when CCE was used as the encoder, and it had something to do with the way older CCE versions handled BFF interlaced input.

But you were talking about the output being still interlaced (i.e. showing combing artifacts) even though you deinterlaced your source. Please clarify if you experienced combing, or if you saw jerky horizontal motion artifacts caused by the wrong field order. Because if you applied deinterlacing then the field order of the output stream does not matter any more.

And it is also a mystery to me how DVDFab HD Decrypter can be able to repair such a stream. Can you please load a VOB before and after DVDFab HD Decrypter into Restream and compare the flags?

Sorry for being too curious...

Cheers
manolito

[spiovente_iv]

...still the whole thing sounds like witchcraft to me, so forgive me for digging...
....
Sorry for being too curious...

No apology needed; you are asking valid questions, and raising valid concerns. To clarify, I experienced combing, but the playback was not jerky.

Here are screenshots of the ReStream analysis, with notes.

First I checked VTS_03_1.VOB from the original source disc, as extracted by DVDFab HD Decrypter 5 with PathPlayer enabled (although the frametype is indicated as being progressive, the video is fully interlaced):

K:\VIDEO\FULLDISC\OLD_SCHOOL_DISC_2\VIDEO_TS\VTS_03_1.VOB

http://img10.imageshack.us/img10/6294/1source.png

Next I processed using DVD-RB 0.98.2(free) with the following project file:
[Options]
Mode=5
OneClick=1
ReduceOpt=0
NoWarn=1
AdditionalOutput=1
LogFile=1
EncoderMinimized=1
RemoveDTS=1
HC_Quality=2
SkinVersion=3
Skin=Rockas Original
HalfD1=0000000
Convert_16_9=0000000
DisableInterlace=1000000
Completed=1
DECOMBFlag=0
idct7Opt=1
VTS_MIN_SIZE=4
VTSM_MIN_SIZE=4
ConvertToYUY2=0
AudioDub=0
AVSFilter01=LoadPlugin("C:\Program Files\AviSynth 2.5\otherplugs\DeGrainMedian.dll")
AVSFilter02=LoadPlugin("C:\Program Files\AviSynth 2.5\otherplugs\SmoothDeinterlacer.dll")
AVSFilter03=SmoothDeinterlace()
AVSFilter04=DeGrainMedian(mode=2)
[Paths]
QuEnc=C:\Program Files\DVD-RB\Encoders\QuEnc\QuEnc.exe
ReJig=C:\Program Files\DVD-RB\Encoders\ReJig\ReJig.exe
HC=C:\Program Files\DVD-RB\Encoders\HC Encoder\HCbatch.EXE
DECOMB=C:\Program Files\AviSynth 2.5\plugins\Decomb.dll
MPEG2DEC=C:\Program Files\DVD-RB\DGDecode.dll
Working=K:\VIDEO\__DVD-RB_WORK\SESAME_STREET_OLD_SCHOOL_DISC_2\
Output=K:\VIDEO\__DVD-RB_OUTPUT\SESAME_STREET_OLD_SCHOOL_DISC_2\
Source=K:\VIDEO\FULLDISC\OLD_SCHOOL_DISC_2\VIDEO_TS\
[Audio]
Selected=11111
[Subpictures]
Selected=


After this step, I removed the original source disc (K:\VIDEO\FULLDISC\OLD_SCHOOL_DISC_2\) from my hard drive, in order to avoid any possibility of mixing up my sources.

Then I eliminated the recursive output directory and checked VTS_03_1.VOB from the DVD-RB output:

K:\VIDEO\__DVD-RB_OUTPUT\SESAME_STREET_OLD_SCHOOL_DISC_2\VIDEO_TS\VTS_03_1.VOB

http://img10.imageshack.us/img10/4438/2dvdrboutput.png

After this step, I used mkisofs to build a DVD-structured ISO from the DVD-RB output, and removed the DVD-RB output files (K:\VIDEO\__DVD-RB_OUTPUT\SESAME_STREET_OLD_SCHOOL_DISC_2\) from my hard drive, in order to avoid any possibility of mixing up my sources.

I loaded the ISO as a virtual drive using DAEMON Tools, and re-extracted via DVDFab HD Decrypter with PathPlayer enabled.

After this step, I removed the ISO from my hard drive, in order to avoid any possibility of mixing up my sources.

... and here is the ReStream analysis of the re-extracted output. Aside from the path (which reflects the title I gave the disc, for DVD-RB output purposes), the properties are all shown as they were for the original source, save that this VOB is truly deinterlaced on playback, and the source was not:

K:\Video\FullDisc\SESAME_STREET_OLD_SCHOOL_DISC_2\VIDEO_TS\VTS_03_1.VOB

http://img10.imageshack.us/img10/5060/3reextraction.png

One thing that is immediately puzzling is that the char string indicates - at this final stage - that the video was encoded by TMPGEnc... which is clearly not the case. Perhaps these are default values to which ReStream reverts, when some other program has removed/modified data, or when the data is not accessible?

Note: I did not alter any settings within ReStream, nor check/uncheck any boxes. I simply loaded VTS_03_1.VOB from each of the respective sources noted, took a screenshot, and exited ReStream.

At this point I am simply glad to have found a workaround that produces what I needed, but like you, I am also curious about why such an odd - and apparently rare - phenomenon would manifest itself in the first place.

- spiv

[manolito]

Hi spiv,

thanks very much for taking the time to analyze what happened. But all I can say is that I have no clue what is going on...:mad:

Looking at your first screen shot already shows that DVDFab HD Decrypter did some bad things to your source during extraction. It shows a VBV buffer size of 4 (illegal for DVD compliant MPEG2), the coded fields have no B-frames at all, and "frametype progressive" plus "progressive sequence" are set which should never happen for an interlaced source.

The second screen shot (after encoding with HC) looks much better. The only irregular thing here is that the first GOP timestamp is not zero.

And the last screen shot (after "correction" by DVDFab HD Decrypter) again shows the same funny values as the first one.


Have you tried to use AnyDVD (or plain old DVD Decrypter in case your source DVD is only CSS protected) instead of DVDFab HD Decrypter?


Cheers
manolito