GOP (IIIIIIIIIIIIII) dvd compatible or not?

[DJRumpy]

SLK, are you sure about that. I could swear I've run across the spec somewhere on the internet.

Funny..while I was entering this reply, I found this page:
http://www.mpeg.org/MPEG/DVD/General/Order.html
which is a branch of this page:
http://www.mpeg.org/MPEG/DVD/

jfebus, what standard are you NTSC or PAL? Also, what player type do you have? I'm wondering if it's just not a GOP length problem?

[jfebus]

jfebus, what standard are you NTSC or PAL? Also, what player type do you have? I'm wondering if it's just not a GOP length problem?

I am using NTSC and a Panasonic V32 DVD Player...I haved tried over 5 different Panasonic DVD Players and still have the artifacts...

Any ideas?

Regards,

[Sonic21]

If you want some interesting reading about Mpeg try this article from Philips

http://www.licensing.philips.com/information/mpeg/documents355.html

or here http://www.mpeg.org/MPEG/DVD/

and a site with all the standards, on all sorts of everything, if you can wade through all the pdf files, here
http://www.ecma-international.org/

Confirm "GOP (IIIIIIIIIIIIII)" is not to DVD standard or compatibility.
If it works you are just lucky with your machine.
It depends on the design of the decoder chip.

[DJRumpy]

jfebus, can you describe these artifacts in more detail? I apologize if you already have, but this thread is getting somewhat long. What type of source is it? (DVD Rip, CAM, VHS, etc..). Also, is it progressive, or interlaced?

[groyal]

In other words, a structure like this one is NOT DVD compliant:

<SEQ>[I][I][I][I][I], <SEQ>[I][I][I][I][I], ...

But a sequence like this one IS:

<SEQ>[I], <SEQ>[I], <SEQ>[I], <SEQ>[I], <SEQ>[I], ...

Unfortunately, this is mis-information, and is simply not true. The DVD spec is a subset of the wider MPEG2 spec. In the DVD spec, it calls out many requirements for a complient stream. Two of them are:
1) GOPs are NOT optional, and
2) Low delay streams are NOT permitted.

Both of the above streams violate both of these requirements, thus neither are complient.

I disagree.

Nothing in the MPEG-2 or DVD standard commands "thou shalt avail thyself of motion estimation at all tymes." So long as the GOP contains an I-frame, a GOP of length=1 is perfectly legal.

Can you provide a reference to a source that says it is not?

But then, I've said enough here. You obviously want to keep asking the question until you get the answer you want to here. Sorry, but life ain't like that. I'm gone...

If I were jfebus, I wouldn't accept "because I said so" as a valid reason either. Why not show us the documentation and let it do the talking?

[jfebus]

Hi DJRumpy,

This is a sample of the artifacts:


I get it using a Vitec DCM MPEG2 Capture Board connected to a VHS.

Regards,

[Sonic21]

I've had that problem!
GOP is irrelivant.
Your are suffering from a combination of jitter, video level and sync problems.
VHS is a pig to author from!

You can cure this with a TBC (Time Base Corrector), expensive.
I overcame it cheaply with a Video Tape Copy Box!
Get a LEK-CPDVD2 under £40 UK

[jfebus]

[quote="SLK001"]

But then, I've said enough here. You obviously want to keep asking the question until you get the answer you want to here. Sorry, but life ain't like that. I'm gone...

No one around is begging you for your Almigthy Book of Knowledge. If you feel you are at a different knowledge level, then don't answer other peoples questions.

Have a nice trip....

[jfebus]

I've had that problem!
GOP is irrelivant.
Your are suffering from jitter and sync problems.
You can cure this with a TBC (Time Base Corrector), expensive.
I overcame it cheaply with a Video Tape Copy Box!

I am getting the error when capturing TV using the VCR connected to my Capture Card! Do you think it's a bad idea?

I will try using a DVD player as a source!

Thanks a lot!

[SLK001]

Nothing in the MPEG-2 or DVD standard commands "thou shalt avail thyself of motion estimation at all tymes." So long as the GOP contains an I-frame, a GOP of length=1 is perfectly legal.

Can you provide a reference to a source that says it is not?

Why do you need me to quote a reference? You seem to have the info at hand already. I have NOT seen the DVD specification - ever. But I have been around doin' this stuff for a while. A lot of good information is available here:

http://members.aol.com/mpucoder/DVD/index.html

There are some people who visit this board who have access to the spec (why the hell this spec is "secret" and requires the "secret handshake" to access is beyond me), but they are not weighing in on this issue (the non-disclosure crap???).

I like to make my Almight Book of Knowledge available to all those who need. But, it looks like I'm goin' on a trip!

A postcard to jfebus:
I'm enjoying my trip! Wish you were here!

[Sonic21]

Great idea that will prove if it is VHS problem.
Make sure you use a non copy/macro protected disc though!

[jfebus]

Great idea that will prove if it is VHS problem.
Make sure you use a non copy/macro protected disc though!

The previous picture was capturing from cable tv using my VCR as a passtrought device to get the video to my Capture Card. Do you think that I need another piece of hardware between the VCR and Capture Card so the Cable Signal can be captured perfectly?

Thanks,

Jose

[Sonic21]

Your idea of trying a capture from DVD is excellent, try that first.
You need to prove if it is a signal problem or capture problem first.

[tomclary]

I like to make my Almight Book of Knowledge available to all those who need. But, it looks like I'm goin' on a trip!

A postcard to jfebus:
I'm enjoying my trip! Wish you were here!

Well SLK001, can you open your Godawful Book of Knowledge and explain what Low_delay means in this context? Your previous posts kept quoting that part of the spec as sufficient argument without ever bothering to say what that means and why it is meaningful. It is sort of hard to get past your arrogance in presentation to actually determine if the information is useful.

[groyal]

Nothing in the MPEG-2 or DVD standard commands "thou shalt avail thyself of motion estimation at all tymes." So long as the GOP contains an I-frame, a GOP of length=1 is perfectly legal.

Can you provide a reference to a source that says it is not?

Why do you need me to quote a reference? You seem to have the info at hand already.

(1) Because I've done my homework with respect to MPEG theory, and
(2) I'm not ashamed to admit when I'm wrong.

You claimed I was spreading "misinformation," and if so, it would benefit me greatly to see the facts for myself. That's why I asked you to provide a reference.

A lot of good information is available here:
http://members.aol.com/mpucoder/DVD/index.html

I agree. But I found nothing there that supports you or contradicts me, so let me give you a hint. All you need is to find a document that reliably establishes one of two things:

(a) That a GOP consisting of a single I-frame is not a valid GOP, or
(b) That the DVD standard prohibits GOPs of length=1.

Either of these will prove that an MPEG-2 bitstream of <SEQ>[I], <SEQ>[I], <SEQ>[I], ... isn't DVD compliant. Not only will I admit I was wrong, I will advise people who've created I-frame-only MPEG streams for editing and archival purposes that the practice is illegal and they should promptly destroy any DVDs they've already burned.

Good luck, and let us know if you come up with anything.

[SLK001]

Well SLK001, can you open your Godawful Book of Knowledge and explain what Low_delay means in this context? Your previous posts kept quoting that part of the spec as sufficient argument without ever bothering to say what that means and why it is meaningful. It is sort of hard to get past your arrogance in presentation to actually determine if the information is useful.

Well, let me try to answer both of youse guys at once. Sorry, but I thought everyone knew what a "LOW DELAY" stream was . Because of the way a MPEG2 decoder handles "B" frames, they are "delayed" from the time they are received until the time they are actually used. Thus, a MPEG2 video stream that is of the "LOW DELAY" type means that there are no "B" frames to have to deal with. For DVD video streams, the "LOW DELAY" is not permitted, thus implying that "B" frames are present.

Now, because of the requirement (or rumor of the requirement) of "B" frames being necessary for a valid DVD stream, I conclude that a GOP of length "1" cannot stand alone for the DVD requirements, because a GOP would have to be a minimum size of two frame ("IB") to meet this portion of the requirement.

Now don't confuse the broader MPEG2 specs GOP requirements (where a GOP of length of 1 is valid, and often used) with the stricter requirements for a DVD. Can I point you to the specific passage in the DVD spec that spells this out? No. Like I said in an earlier post, I haven't seen it and I doubt that it will ever be available on the net. That, coupled with the fact that I will never sign any non-disclosure agreement, let alone shell out $5,000 US for the docs make me a "rumor monger". All the info I have on the spec comes basically from rumors, backed up by empirical evidence (but that are basically still rumors).

But what the hell... If an all "I" sequence is the only thing that works for you, then use it. Since most DVD players have the ability to go "beyond" the full DVD complience requirements, you are not likely to suffer much from the occassional "strict" player.

But I have to create DVDs that are compatible with ALL players, so I have to be more careful, and stick to the spec (as I have been able to gleen from my research).

[groyal]

Well, let me try to answer both of youse guys at once. Sorry, but I thought everyone knew what a "LOW DELAY" stream was :wink: . Because of the way a MPEG2 decoder handles "B" frames, they are "delayed" from the time they are received until the time they are actually used. Thus, a MPEG2 video stream that is of the "LOW DELAY" type means that there are no "B" frames to have to deal with. For DVD video streams, the "LOW DELAY" is not permitted, thus implying that "B" frames are present.

I'm afraid that's not correct. A "low delay" MPEG stream is one that's optimized for streaming, such as real-time display over the web. It doesn't use B-frames, but that's not the reason it's prohibited by the DVD standard: it's illegal because it changes the way the pictures in the stream are interpreted.

Now, because of the requirement (or rumor of the requirement) of "B" frames being necessary for a valid DVD stream, I conclude that a GOP of length "1" cannot stand alone for the DVD requirements, because a GOP would have to be a minimum size of two frame ("IB") to meet this portion of the requirement.

You're assuming that DVD requires B-frames because low-delay streams don't make use of them; that's not a conclusion that can logically be drawn from the facts. B-frames are bidirectionally estimated, so they can't be present without at least one other frame to refer to: "I", "IP" and "IBP" and "IBI" are legal GOPs, but "IB" is not.

Now don't confuse the broader MPEG2 specs GOP requirements (where a GOP of length of 1 is valid, and often used) with the stricter requirements for a DVD.

On the contrary: I think you're confused about what constitutes a valid GOP, which has nothing to do with either specification per se. An I-frame-only MPEG stream is unusual, but it is not illegal. If it were, it should be easy to find some kind of documentation that describes this limitation, don't you think?

Can I point you to the specific passage in the DVD spec that spells this out? No. Like I said in an earlier post, I haven't seen it and I doubt that it will ever be available on the net.

That's part of my point: you're not going to find a document that spells this out because no such document exists. That's why it's so important to double-check yourself before accusing anyone of spreading "misinformation": if you happen to be mistaken, you leave people with the impression that you know less about the subject than you'd like them to believe.

[Emiel Schiphorst]

A bit late(r) (and extending this thread), but maybe interesting ...
a possible alternative.

I'm working on producing MPEG EX (a proprietary MPEG-1 from Sony) by software. In contrast to what most people think, this is not MJEG or Motion-JPEG, but compliant MPEG-1.

This MPEG-1 has an IPP GOP, with dummy/empty P-frames (an extreme case: ZERO motion estimation). These 2 dummies consume bytes though. Viewing the 3 frames one after the other shows no difference at all!