Upsides/downsides to open GOPs

[dadrab]

Prompted by another poster here and fueled by the fact that DVDLabPro always warns me that I have open GOPs, I thought I'd ask the question:

What difference does it make?

The discs I burn always work in all my machines and I never get complaints from folks whom I give discs to.

I must admit that I always ignore the warnings DVDLabPro gives me, but it's because I've never run into trouble.

And that leads nicely into the second part of the question: How do I get closed GOPs? The authoring app. gives me the warnings whether I feed it video that I've encoded or video that I've captured via my PVR-350.

Thanks all.

[redwudz]

From our Glossary for 'GOP':

The decoding process, as the GOP header is immediately followed by an Intra picture, can begin at that point of the bitstream. Anyway it's possible that some B pictures, following such I_picture in the bitstream, have references coming from the previous GOP and can't be correctly decoded.
In this case the GOP is called an Open GOP because some references from the previous GOP exist; if a random access to such a GOP is performed, some B_pictures shouldn't be displayed .
A GOP is called a Closed GOP when either there are no B_pictures immediately following the first I_picture or such B_pictures haven't any references coming from the previous GOP (in this case a GOP header flag must be set).

Someone else will have to answer your questions, but that's what the differences are.

[Midzuki]

https://forum.videohelp.com/topic347269.html

[dadrab]

https://forum.videohelp.com/topic347269.html

Yeah, this was the post that prompted the question. I know you can turn it off, but do I want to?

[Midzuki]

How do I get closed GOPs?

If you meant "how to close the GOPs after the video compression has been done",
then my answer is "I don't know, and I don't want to know".

[guns1inger]

Pro for open GOPs - slightly better compression

Con (according to DLP) is less accurate chapter placement. However, as true chapter placement only occurs on an I frame, I suspect this is really a furphy. If you want truly accurate (frame accurate) chapter placement then you have to encode with I frames where you want your chapters to be. Most decent encoders have a facility, either through text file or in-program selection, to do this.

[manono]

There's another con and that is if you plan on editing the result, like joining encoded clips together, you may find artifacts at the join unless you use closed GOPs. Also, I believe angles require closed GOPs. But personally, I only ever use Open GOPs for the same reason you gave - slightly better quality/compression for the same file size.

[Nelson37]

Learn something new every day.

I thought an I-frame had to be a complete picture, and that all references went forward from there. Why would they reference before a complete picture?

Also thought Open GOP meant it could have a variable number of frames after the I-frame. Is there another term for variable GOP length?

Closed edit better, also seem to have less audio issues, these may be related. All my stuff is captured, this also probably affects this.

[guns1inger]

In an ideal world you would be editing before encoding to mpeg-2.

[manono]

In an ideal world you would be editing before encoding to mpeg-2.

Is that for me? I'll give you some examples of what I mean. I'm reencoding a DVD and want to give the end credits lower quality than I do the main movie, in order to free up bits to improve the quality of the main movie. I encode them separately and join the 2 parts during authoring. Now, credits often begin with some black frames before you get to the typical scrolling white on black credits. I can keep Open GOP because you won't get artifacts (in the form of different colored blocks) when joining black-on-black. But not all end credits are like that and encoding with Closed GOP might help. Or, say, I want to do the same with the opening credits - encode them for a lower quality. They usually won't have black frames before going into the main movie, so there's a case where encoding everything with closed GOP would definitely help. Or I'm encoding an anime and the opening song is progressive 29.97fps, or the end credits are interlaced 29.97fps, but the main part of the episode is 23.976 fps hard or soft telecined film. I might (and have often done this) encode the different parts separately and rejoin them during authoring. Encoding using Closed GOPs is a must, unless you want 2 frames of junk at the joins.

Maybe those things aren't defined as editing, but I think joining clips is editing.

[Abond]

All the MPEG Encoders with proper scene change detection are closing the GOPs at scene change and at the beginning of the clip (first GOP). For CCE, TMPGEnc, HC - I am sure they do (checked). That is your stream is mix of closed and opened GOPs. Unfortunately all libavcodec (or ffmpeg) based encoders don't have proper scene detection and consequently don't close GOP's at scene change, but only at the clip start (even for this I am not sure). At least they don't flag the header.
https://forum.videohelp.com/topic328975.html#1700104

[hank315]

I might (and have often done this) encode the different parts separately and rejoin them during authoring. Encoding using Closed GOPs is a must, unless you want 2 frames of junk at the joins.

If you have two separately encoded clips you can always join them regardless if the GOPs are open or closed.
This is because the first GOP is always closed.

[dadrab]

In an ideal world you would be editing before encoding to mpeg-2.

Unless capturing directly to MPEG2 as we PVR-150, 250, 350 users must do. So, the editing is then done in VideoReDo or some other like application that works within that format.

There's another con and that is if you plan on editing the result, like joining encoded clips together, you may find artifacts at the join unless you use closed GOPs.

Hmmmm. Very interesting.

Learn something new every day.

Ain't that the truth.




So, for the folks who capture directly to MPEG2, how do we get closed GOPs without re-encoding? Or do we?Obviously, if we capture at an unusually high bit rate, we can re-encode down to one within DVD spec. and close the GOPs.

Recommended or not?

[GeorgeW]

If you are doing multiple angles, then DLP recommends Closed GOP's (to make the angle changes better).

Regards,
George

[manono]

I might (and have often done this) encode the different parts separately and rejoin them during authoring. Encoding using Closed GOPs is a must, unless you want 2 frames of junk at the joins.

If you have two separately encoded clips you can always join them regardless if the GOPs are open or closed.
This is because the first GOP is always closed.

Hi hank315. Nice seeing you here.

I didn't say they wouldn't join. I said you can often get 2 frames of junk (in the form of brightly colored blocks) at the join.

[hank315]

Hi manono,

Do some occasional reading here also

About the joining, if the first GOP of the second clip starts with IPB... (stream order), it should work OK, then the GOP is closed.
The brightly colored blocks mean of course the first B-frames of the second clip try to reference the last P-frame from the previous GOP, then the second clip starts with an open GOP, IBBPBB...
But if the second clip isn't edited, (cutting of the first closed GOP), IMHO it shouldn't produce junk frames.

If you join the clips do you also remove the sequence_end_code header from the first clip (four last bytes: 00 00 01 B7)?
Some HW decoders don't like it when there's a sequence_end_code in the middle of the stream.

[manono]

If you join the clips do you also remove the sequence_end_code header from the first clip (four last bytes: 00 00 01 B7)?

No, I don't know anything about that kind of thing.

Here's what I remember doing once. I had come out a little bit oversized. Rather than reencode the whole thing, I opened the final reencoded (Open GOP) DVD in VobBlanker and extracted the first cell. I reencoded it for a slightly smaller size and replaced that same first cell with the slightly smaller sized one. There were 2 artifacted and blocky frames at the join. That first cell didn't end at a scene change, but was right in the middle of a scene.

This was a while back, so I can't remember all the gory details. Maybe I extracted the cell using PGCDemux, but I don't think so. Probably wouldn't make any difference anyway.