Progressive (deinterlaced) video stream flag set to interlaced at rebuild [Archive]

Observer

24th July 2004, 20:04

DVD-RB 055b set the video stream flag of a de-interlaced movie to interlaced at rebuild.

If I start with an Interlaced source and select the option to Deinterlace the movie. Then the created m2v-files are progressive after encoding with CCE. However at the rebuild stage DVD-RB seems to set the video stream flag back to Interlaced (as the original was)which is probably wrong since the video streams are progressive after deinterlacing and progressive encoding in CCE.

Observer

25th July 2004, 21:22

De-interlacing and some filtering is often necessary to get a decent result with a certain type of movies.
Therefore it would be nice to get some clarification on why the movie stream flag is set to interlaced during the rebuild process when the movie has been de-interlaced and encoded as progressive. I'm afraid that this may create playback problems on some DVD players.

Joergen

25th July 2004, 21:26

Are you sure the disable interlaced "tick" is still active when you rebuild? Not so long ago I encoded a movie with disable interlaced and the output vobs were flagged progressive as they should, but you might have to remember to retick the options if you close RB.

Observer

25th July 2004, 21:35

I will try to rebuild one of the movies aigain and let you know but this is what I have seen so far.

The original movie is truely interlaced.
The encoded m2v files are flagged progressive.
The VOBs after rebuild are flagged interlaced.

Joergen

25th July 2004, 23:08

Looks like you're right. I checked a couple movies I had done with the same scenario as you have above where movie is interlaced, "disable deinterlaced" is chosen, m2v's are progressive, and the rebuilt vob's are again wrongly marked interlaced. This is with PAL material, which prolly is more safe in this scenario.

Perhaps that can cause weird playback (sync, dropout) errors with some mpeg2 decoders and NTSC content, although most people probably dont use deinterlace options.

Observer

26th July 2004, 19:44

It doesn't seem matter what settings I use. During rebuild DVD-RB seems to put the video stream flags back to what the original source was.

frank

27th July 2004, 11:32

It doesn't seem matter what settings I use. During rebuild DVD-RB seems to put the video stream flags back to what the original source was.Yes, its a bug I described a long time before. But nobody heard my crying...
DVD-RB 0.49 and progressive encodings (http://forum.doom9.org/showthread.php?s=&threadid=76906)

barneypooch

1st August 2004, 11:08

Any more news on this? Or any ideas for a workaround? My DVD player is flag-reading and this bug has effectively stopped me using RB.

Chibi Jasmin

3rd August 2004, 15:40

You know a reliable way to find out, if your dvd-player reads the interlaced/progressive flag from the stream (and if it adjusts decoding according to that)?

Maybe there's some test patterns one could encode and try available that make it easy to see without any special tricks?

TheSeeker

3rd August 2004, 17:42

I noticed this as well. Im deinterlacing Taken miniseries which is truly interlaced and i figured i would tick the disable interlaced option when i went to prepare so it wouldnt be there in the avs script in the converttoyuv line. But if you tick this option before prepare it disables any deinterlacing. So im not sure if i have to tick this only when im rebuilding and leave it unticked when i prepare and encode? Not really sure how to procede but i would definately like truly progressive dvd's for when i get a progressive setup.

barneypooch

3rd August 2004, 18:59

I'm surprised more people aren't picking this up. If (like me) you have a flag-reading DVD player, this is a big problem. Are we posting in the wrong place?
At the moment I can't use RB because of this.
Any suggestions from the smart guys? I never asked a question before and I did make a (PayPal) contribution... :)

TheSeeker

3rd August 2004, 19:33

could you just deinterlace the content and have the disable interlaced unticked in the prepare and encode phases then when it comes time to rebuild could u just go and tick that option then rebuild? would that work? anyone care to try?

jdobbs

3rd August 2004, 20:00

Just click on "Disable Interlaced" and the source will be treated as progressive. Just to make it easier, I'll make the progressive flag automatic when you select "Deinterlace with Decomb" in the next version. The reason it is manual now is so you can select it if you use some other filtering to deinterlace...

It isn't a bug, it's a feature. ;)

TheSeeker

3rd August 2004, 20:13

@jdobbs

I just thought it was wierd that when i had both deinterlace with decomb and disable interlaced both checked and i did a prepare, it did not put the field deinterlace in my avs scripts. but your saying just choose disable interlaced when i do rebuild right?

jdobbs

3rd August 2004, 20:21

Yes, and it used to work. But I'll check tonight and make sure I haven't screwed something up. I may have somehow changed the order of precedence.

Just a note to everyone... most of the time you will get a better picture by leaving a source interlaced rather than doing deinterlacing. This option should be used with discretion... if you are deinterlacing your sources on a regular basis (not just special cases) you probably aren't getting an optimal picture.

I realize some folks watch movies on their PCs and some interlaced sources look bad on monitors...

Joergen

3rd August 2004, 20:47

Dont forget that us PAL users like to just disable deinterlace in order to encode progressive (and so get a theoretical increase in quality from CCE) since the frames look progressive but are encoded interlaced in the originals often.

barneypooch

3rd August 2004, 21:03

JDobbs - something has really changed in recent versions of RB. About three builds ago (sorry, can't now remember exactly), if you checked "Disable Interlaced", the resulting ECL file would have the "progressive" set to "1". Which it still does. But back then, the final VOB files would also have the progressive franes flag set to yes.
But now, during the rebuild phase, the "progressive frames" flag in the final VOB files gets turned off somehow, so that a flag-reading DVD player thinks it's dealing with interlaced and screws up.
This is quite an issue for those of us in PAL-land with projectors or progressive-capable TVs. God only knows why the studios can't set the progressive flag correctly, as they do in NTSC world. But they don't.
Any answers appreciated.

Eki

3rd August 2004, 21:42

Originally posted by barneypooch
God only knows why the studios can't set the progressive flag correctly, as they do in NTSC world. But they don't.
Any answers appreciated.

Studios do get it right most of the time. Frame type flag only tells you which encoding mode was used. It is very common that progressive PAL content is encoded in interlaced mode. Encoding progressive content in interlaced mode makes it actually interlaced so the flag is set correctly.

TheSeeker

3rd August 2004, 21:48

The question then becomes why the sam hell anyone would want to encode truly progressive content as interlaced?

barneypooch

3rd August 2004, 22:13

Eki, I'm sure you're trying to help me but I'm too stupid to understand!

Are you saying that the "progressive frames" flag makes no difference to a flag-reading DVD player? If so, then what flag, exactly, does the machine read in order for it to turn on its de-interlacing circuits? And why do I get de-interlacing artifacts on progressive disks which are incorrectly flagged and yet, if I use ReStream to change the "progressive frames" flag, the same disks play without artifacts?

Anyways, all I'd like is for RB to stop changing the progressive frames flag when it gets to the rebuild phase. Just like it (didn't) used to.

jdobbs

3rd August 2004, 22:14

Originally posted by barneypooch
JDobbs - something has really changed in recent versions of RB. About three builds ago (sorry, can't now remember exactly), if you checked "Disable Interlaced", the resulting ECL file would have the "progressive" set to "1". Which it still does. But back then, the final VOB files would also have the progressive franes flag set to yes.
But now, during the rebuild phase, the "progressive frames" flag in the final VOB files gets turned off somehow, so that a flag-reading DVD player thinks it's dealing with interlaced and screws up.
This is quite an issue for those of us in PAL-land with projectors or progressive-capable TVs. God only knows why the studios can't set the progressive flag correctly, as they do in NTSC world. But they don't.
Any answers appreciated. I really haven't changed anything in this area. I do these all the time and haven't had a problem. If you set "Disable Interlaced" the source is treated as progressive and stays that way. The VOB files would also be set progressive. Check the REBUILDER.INF file and see if "progressive" is set to "1" in the appropriate segment(s).

Like I said, I'll check it tonight and make sure it's okay...

jdobbs

3rd August 2004, 22:19

Originally posted by Eki
Studios do get it right most of the time. Frame type flag only tells you which encoding mode was used. It is very common that progressive PAL content is encoded in interlaced mode. Encoding progressive content in interlaced mode makes it actually interlaced so the flag is set correctly. Sometimes you'll even see an obviously progressive source (no combing at all) actually split into seperate fields in the MPEG stream, not just flagged. I don't think the quality would suffer from that on a "normal" TV. But you'd never get that rock solid progressive display.

Observer

3rd August 2004, 22:34

I've tested it one more time and made sure I was checking "Disable Interlaced" but the problem I originally described is still there. The video stream flag is set back to Interlaced during rebuild.

The original movie is truely interlaced.
The encoded m2v files are encoded and flagged progressive.
The VOBs after rebuild are flagged interlaced.

Why is the video stream flag changed during rebuild?

mrslacker

3rd August 2004, 23:00

Originally posted by jdobbs
Just a note to everyone... most of the time you will get a better picture by leaving a source interlaced rather than doing deinterlacing. This option should be used with discretion... if you are deinterlacing your sources on a regular basis (not just special cases) you probably aren't getting an optimal picture.

I have found the same in my experience. Just let your player deinterlace! If it claims 3-2 detection and etc., it should properly deinterlacing film based material as well. A lot of players have methods more effective than field deinterlacing, such as motion adaptive deinterlacing. Also, for those with flag-reading players rather than cadence-reading players, it would seem much safer to not deinterlace as your player counts on correct flags to describe the video... Your player could drop to video mode deinterlacing or not deal at all if the proper flags are not set for the actual encoding sequence. This is clearly a possibility with DVD-RB or any other "reauthoring" application. However, There are already enough quirks in "professionally" authored discs that require a player that can detect changes in cadence and correctly deinterlace. A purely progressive disc could effectively solve these issues at the risk of creating more. I strictly leave the source format alone unless I'm trying to experiment.

mrslacker

3rd August 2004, 23:27

Originally posted by barneypooch
Eki, I'm sure you're trying to help me but I'm too stupid to understand!

Are you saying that the "progressive frames" flag makes no difference to a flag-reading DVD player? If so, then what flag, exactly, does the machine read in order for it to turn on its de-interlacing circuits? And why do I get de-interlacing artifacts on progressive disks which are incorrectly flagged and yet, if I use ReStream to change the "progressive frames" flag, the same disks play without artifacts?

The progressive_frame MPEG flag indeed tells the deinterlace chip to work on flag-reading players. picture_structure also plays a roll in describing the sequence, but this probably won't play a roll in this case. Eki is saying that just because material is encoded as interlaced doesn't mean it's not progressive content. However, in your case, the material is definitely not encoded interlaced (you are using FieldDeinterlace), but the flags haven't been set.

Here's what's happening to you with no progressive flags set:
The player is combining fields that are stored in one MPEG frame, but came from two different film frames, causing combing. Get a cadence detecting player. :(

Observer

4th August 2004, 00:05

barneypooch

4th August 2004, 08:48

Observer:

The original movie is truely interlaced.
The encoded m2v files are encoded and flagged progressive.
The VOBs after rebuild are flagged interlaced.

Why is the video stream flag changed during rebuild?

Thanks for putting it more clearly and concisely than me. That is exactly it.

MrSlacker: Oh, how I wish I could get a cadence-reading player! But I use DVI to connect to my projector and the only player that gets the job done without a $2,000 price tag right now is the Momitsu 880, which is based on a Sigma Designs chip and is flag-reading. The Samsung 931 has DVI and a nice Faroudja chip but sadly a bunch of other problems which make it next to useless.

However, if the flags are set correctly on a progressive disk, the 880 works like a charm. Go over to avsforum and you'll see how this issue has been a subject of hot debate for months now.

barneypooch

4th August 2004, 08:50

The original movie is truely interlaced.
The encoded m2v files are encoded and flagged progressive.
The VOBs after rebuild are flagged interlaced.

Why is the video stream flag changed during rebuild?

Thanks for putting it more clearly and concisely than me. That is exactly it.

MrSlacker: Oh, how I wish I could get a cadence-reading player! But I use DVI to connect to my projector and the only player that gets the job done without a $2,000 price tag right now is the Momitsu 880, which is based on a Sigma Designs chip and is flag-reading. The Samsung 931 has DVI and a nice Faroudja chip but sadly a bunch of other problems which make it next to useless.

However, if the flags are set correctly on a progressive disk, the 880 works like a charm. Go over to avsforum and you'll see how this issue has been a subject of hot debate for months now.

barneypooch

4th August 2004, 09:28

barneypooch

13th August 2004, 11:05

Sorry - I don't know how that got posted 3 times!

Any definitive answer to this question?

jdobbs

13th August 2004, 18:58

Yes. It was a bug. It has been fixed and will be in the next version. I still suggest, however, that the DECOMB function be used very rarely. It normally isn't a good thing.

barneypooch

13th August 2004, 19:03

Thank you very kindly indeed, Mr D!

I would never use decomb. My only problem was with the progressive flag getting set wrong and screwing up my flag-reading DVD player.

Thanks again. Does this mean I have to make another contribution :D

jdobbs

13th August 2004, 19:27

Hmmm... the only time the flags would be set wrong would be if you had DECOMB selected... because it would result in a progressive output from an interlaced source. The bug was cause by the sequence in which settings were being checked.

Nope. One donation is plenty. If everyone who uses DVD-RB donated, I'd do it fulltime and it would have been out of beta long ago. But as it is, you are already part of the .1% who pushed the button...:)

Chibi Jasmin

18th August 2004, 14:22

Well, a problem I often encounter is the following...

Source: PAL DVD progressive content, flagged interlaced

Looks fine, when decoded with interlaced=false (interlaced=true gives typical stripes in bright colors etc.), so I set DVDRebuilder to DisableInterlaced.

DVDRebuilder sets up AVS Scripts to encode progressive/zigzag in CCE, so far so good.

BUT on rebuilding it puts in the new prog/zigzag stream with interlaced flags. :(

What I did so far to circumvent that is to still set disable interlaced in rebuilder, but manually edit the ecl files to uncheck progressive and use alternate scan to that on remux stream and interlaced flags fit together. Would be nicer to have DVDRebuilder just set the correct progressive flags, then I could encode with prog/zigzag which should be better for progressive content.

jdobbs

18th August 2004, 23:31

BUT on rebuilding it puts in the new prog/zigzag stream with interlaced flags. I think you are mistaken. I've checked the code, and I will check it again to be sure. But setting "Disable Interlaced" forces progressive and the interlaced flags are not set on rebuild. Try it again and ensure you truly have "Disable Interlaced" set -- remember this flag is not saved to project files so it cannot be used in batch mode...

Observer

19th August 2004, 07:07

@jdobbs
But it doesn't work correctly. Have you tested it on a PAL DVD?
I've tried the option "disable interlaced (apply to all)" on at least five different movies and it doesn't matter. The stream flags are set back to interlaced during rebuild.
This is actually what both me and Joergen reported at the beginning of this thread.

Regards,
O

Chibi Jasmin

19th August 2004, 16:17

Have checked that many times...I stick to what I said :D

Just to say again, I am not talking about deinterlacing with decomb!

I am talking about setting "disable interlaced" in avs settings...stream gets correctly encoded zigzag/progressive, but muxed with interlaced flags (as they were in original).

djan

19th August 2004, 22:37

Hi,

Same problem here with DVD-RB 0.56. Not talking about deinterlacing or anything else. Just talking about asking to DVD-RB to treat the source as progressive and to set the flag as progressive. But it doesn't work. The flag at the end is always set to Interlaced. Just checked using DVD2AVI. It has to indicate it's progressive and not Interlaced. Please Jdobb, check it to see if it's not a bug. Thank you !

jdobbs

20th August 2004, 03:35

I'll find it and fix it.

Chibi Jasmin

20th August 2004, 12:28

Thanks a lot! Looking forward to new version...

Chibi Jasmin

23rd August 2004, 18:02

Any progress on finding and fixing the problem?

djan

23rd August 2004, 18:54

Waiting too :)

Blackout

23rd August 2004, 22:48

Further, just wanted to request that the "Disable Interlaced" settings be actually saved with the project...so if you try and do any batch processing and one of the dvds needs the progressive flag forced, it does reload your setting for processing....

Blackout1230

Chibi Jasmin

27th August 2004, 14:39

*bump*

jdobbs

29th August 2004, 21:23

Well... I stand corrected. I found what was causing this error -- and it has been fixed. It will be included in v0.57. I'd misspelled the variable name which meant its "Disable Interlaced" condition was never true when setting the progressive flag in the .FLG file.