Different GOP settings affect clarity of final video?

[joeAgain------]

BJ_M has told me that lowering my GOP settings during capture will make editing and processing of the video files easier(on the software, I guess).

My capturing software's default GOP settings are an I Frame distance of 15, with a P Frame distance of 3(and he suggested an I Frame distance of 4, with a P distance of 1).

My GOP settings are adjustable and that sounds fine to me(I'd be happy to make the processing easier and quicker for my editing and authoring programs, because that sounds like a good thing)...but I'm curious about if different GOP settings affect the clarity and quality of the final video image after burning and viewing on my DVD player/TV.

Can somebody tell me about this, please?(BJ_M, or anybody?)

[ronnylov]

Some editiong software, like the editor in TMPGenc can only cut on I-frames. Also DVD Authoring programs can only set chapter points on I-frames. So a shorter distance between the I-frames makes the precision of the editing better. Some editors like M2-Edit or Womble MPEG2VCR can make the cuts on all frames. The editor reencodes the actual GOP structure to make the cut frame an I-frame.

But on the other hand, I-frames requre more bitrate than P-frames and B-frames require less bitrate than P-frames. So if your video has a low bitrate it may look better with less I-frames and more B and P frames. If you capture at a very high bitrate you may get best results with I-frames only. You have to do some tests and see what GOP structure you like best at your bitrate and resolution. It also depends if there is a lot of action in the video or if it's only low motion video.

[joeAgain------]

Thanks, Ronnylov.
That helps me understand a little more. I nearly always capture pretty high at between 2Mbps and 8Mbps (and mainly use Womble to trim my Mpeg-2 caps from my Dazzle system).

Could you explain a little more about the amount of action in the video being affected by the number of I Frames? I guess you are saying that more I frames is clearer in action video, the same way a high bitrate would? (or have I got it backwards about I frames and action?)

[flaninacupboard]

right, an I frame is to all intents and purposes like a jpeg image. you can open it in media player or and editor of some kind and there it will be, it's an image. a P frame however is the difference from the last frame, a P frame can only exist in a chain with an I frame at the start, as it only records the -differences- between the two. P frames require much less data than an I frame, as it's not a complete image. however, because it's not a complete image, it's subject to image artifacts. (blocks, smears, inaccurate colour depth, banding and solarisation etc etc etc) a B frame is even more compressed than a P frame, and records the changes both from the last frame and to other frames in the GOP. again B frames are open to all kinds of artifacts. so in a typical VCD GOP of IBBPBBPBBPBBPBBPBB one frame is complete and you've got 17 highly compressed not as good looking frames. the problem of editing also applies, you can only make a cut or join every half second. if you alter your GOP so it goes IIIIIIIII then image quality is better -if- your bitrate is high enough. i think the only time this is acceptable is if you're capping at 15mbps, the maximum MP:ML Mpeg2 will allow (i believe) if you want to capture lower than this (you said 8mbps as an example) then something like IBPB gives you a great mix of editability compression and quality.
*phew!*
although what are you doing with these mpegs? are they being re-encoded?

[joeAgain------]

Thanks, Flaninacupboard.
That helped me.

Well my Dazzle system is only capable of capping into Mpeg. So I cap our family VHS home videos, do minor trimming(with Womble), author with chapters, and burn onto a DVD-R to preserve the memories. (I'm trying to preserve 15 yrs of videos before they degrade and the tapes wear out...which is already starting to happen)

I just recently became interested in inserting fun effects between segmants with certain authoring programs. My GOP question aroze because I was told that editing my raw video before burning could be done easier with more I Frames than I had been capping with before(to this point I've been using the same structure as in your example...IBBPBBPBBPBBPBBPBB). The suggestion was that IPPP was better.

What would be the difference between IPPP and your great mix suggestion of IBPB? And would there be any perceptable difference in picture/image quality on my DVD between those two?
(Man!...I am Really into new territory on this topic for me) I appreciate your thoughts.

[BJ_M]

for editing -- we try to avoid B frames (when capturing) - for a couple reasons ..

one - you must understand mpeg's are not written in a linear fashion with IPB frames .. but are with IPPP frames ..
and there is not much "there" with a B frame (this doesnt mean they are bad and you shouldnt use them in your final render) ...

the short GOP for capturing does assume higher bit rates for capturing - 8meg/s and higher is fine for IPPP

[joeAgain------]

Thanks, BJ_M.
Guys, would there be a good comprimise structure for 2 or 5Mbps?

You said,"(this doesnt mean they are bad and you shouldnt use them in your final render) ...", but I'm still not sure what you mean by "final rendering". I only intend on intentionally setting or changing my GOP for the capture....in fact I don't know where else I have options to do it. I never re-encode with TMPGEnc, if that is what you are talking about since I'm already in mpeg-2 format at capture.

I never cap higher than 8Mbps...and am usually a little lower, since it burns up too much DVD-R disk space...and in light of the VHS video source being of marginal quality anyway, that always seems like a wasted effort for no appreciable quality gain to go extremely high, you know?.

[joeAgain------]

Are you guys still there?

I'd really like to know if there is a good shortened structure for 2 or 5mbps Mpeg-2 captures.....and BJ_M, in my authoring programs (PowerDirector and MovieFactory2) just were would I reset the GOP structure for the final rendering, as you put it?

(again, I don't need or want to re-encode with TMPGEnc., besides the fact that it would take "days" to do that on my 4.5Gig video files)

[flaninacupboard]

fair enough, i didn't realise B frames caused a problem with editing
personally i cap in DV so only worry about GOPs on output. the best structure for 2~5mbps is the standard IBBPBBPBBPBBPBBPBB. if you're after better quality, then i really reccomend you either cap 15mbps I only or 10mbps IPPP and re-encode to whatever bit rate you like using IBBPBBPBBPBBPBBPBB for your output, as this is DVD standard and the best trade off you get between quality and compression. yes, tmpgenc is slow at encoding, CCE is trickier to start using and actually costs money (unless you look in the right place) but is so much quicker than tmpgenc, and produces quality if not the same then better. personally the quality is the same, the speed is a huge imrovement.
Really, MPEG is not the best format for captruing and editing or archiving. it really should only be used as a final output, that's what it's designed for. Capping at very high bitrates and re-encoding will give you a huge improvement in quality, and also enables you to know exactly how much you can fit on a disc, because you control the edit to the frame and the bitrate to the kilobit.

[joeAgain------]

Thanks flaninacupboard,

I thought since I have so many years of analog VHS tapes to copy that my choices were pretty limited on capture systems. I've been pretty happy with my Dazzle system, but wish I had a nickel for every time someone has told me that capturing in DV/AVI is much better....HeHe

And I know it's true, but have to deal with what I've got. I've never had any problem editing with Womble.

Thanks....I'll look into CCE if I ever want to spend the time to re-encode, but I'm kind of glad I don't need to with my Dazzle Mpeg system.

Thanks again

[joeAgain------]

I guess I need to say this again,
I am ONLY able to work in mpeg format because that is ALL that my Dazzle system captures in.

Thanks for the thought of CCE, but I downloaded a trial version and doesn't import or re-encode mpeg....only DV.

Thanks though.





,

[disturbed1]

If the only editing your doing is simple cuts, the default GOP structure is just fine.

The Dazzle captures with closed GOPs, so any compitent mpeg editor won't have an issue with these files. Such as womble. Your GOP looks like this IBBPBBPBBPBBPBB IBBPBBPBBPBBPBB. Or two GOPs per second. This allows for 1/2 second (15 frame) editing. If this doesn't produce good enough results, you can always add in a fade at the end to hide the unwanted 15 frames.

With this GOP structure, it ensures the greatest compatiblity for DVD authoring. The recomended GOP structure for NTSC (according to MPEG) is exactly what the DVCII does. GOP shoud be equal to one half FPS, or a GOP length of 15 for NTSC, and a GOP lenght of 12 for PAL. DVD has a maxium GOP length of 18 for NTSC, and 15 for PAL, with no more than 2 B frames between P frames.

At lower bit rates increasing the distance between I frames will increase quality. Decreasing the GOP, decreases the quality for a given bitrate.

http://www.ece.concordia.ca/~sorial/seminar1/node24.html

When capping for SVCD through my DVCII, I use a GOP distance of 33. This helps decrease most of the blockiness caused by low bitrates.

Ligos has articles about this, and recomends the same. TMPG also modified it's templates for SVCD sometime ago with similar settings.

There are countless articles writen by university professors that will most likely provide more information than needed. If your interested, spend a few hours searching with google

[ronnylov]

If you edit with Womble or M2-Edit then you can cut on each type of frame becuase this editor will reencode the GOP at the cut point (but leave the rest untouched). It's a frame accurate editor so you don't need to worry much about this.

What resolution do you capture with? It's hard to know if the bitrate is enough if you don't mention the resolution. Since you're nor reencoding (and I do not recommend reencoding either) and are capturing from old VHS source I would recommend half D1 resolution (352x576 or 352x480 PAL/NTSC) and a bitrate of 4 Mbit/s. If you capture at full D1 (720 width) then I recommend 8 Mbit/s. But I don't think you'll gain any quality with full resolution from a VHS source. The half D1 resolution is a DVD compatible resolution and is a good choise for VHS source. Don't go above 8 Mbit/s if you want it DVD compatible.

It's hard to recommend any specific GOP structure. You have to try and see what results you get. Perhaps something like IBPBPB or IPPP or even IBBPBBPBBPBB may be worth a try if the defualt GOP is too blurry in high motion scenes. The default GOP is probably best at bitrates below 3 Mbit/s with half D1 reolution. I don't recommend a bitrate below 3 Mbit/s when capturing home videos at half D1.

If your video is shaky then I recommend to try DynaPel Steadyhand antishake software. But this only works on avi files so it's prbably not so good if you capture in mpeg2. It's great for DV avi files!

Good Luck!