How does GOP size affect recording ?

[Serville]

I don't know what GOP size is, but many people suggest to use GOP size 12 for PAL recording. The default GOP in my WinTV2000 (with my Hauppage USB2 PVR2) is 15. How will this affect recording ?

[guns1inger]

The concept of a GOP, or Group Of Pictures, as it applies to mpeg encoding, is explained in this article -> http://en.wikipedia.org/wiki/MPEG-2

For PAL the maximum length of a GOP is 15 frames, for NTSC is it 18. Longer GOPs can provide better compression.

[ron spencer]

I use 15 for ntsc....this gives chapters every 29.97/15=0.5 seconds....which is perfect. that is I frame every 0.5 sec and chap can only occur on I frame

[guns1inger]

Those numbers are a maximum for the format, not absolute. So 15 is perfectly legal for NTSC, as 12 is for PAL.

[jman98]

Those numbers are a maximum for the format, not absolute. So 15 is perfectly legal for NTSC, as 12 is for PAL.

Actually, the maximum values are 18 for NTSC and 15 for PAL. See the Wikipedia article mentioned above (look towards the bottom) to confirm this. 15 for NTSC and 12 for PAL are the usual values and some authoring applications incorrectly enforce these as upper limits. I had a problem once where I recorded a laserdisc to DVD format and one GOP had 16 frames for some reason and there was nothing I could do to change it. I tried everything. I even tried re-recording the same side of the laserdisc and I still got a 16 frame GOP in exactly the same spot. DVD Maestro just freaked out at this one GOP of 16 frames and refused to make the DVD. That made me learn how to use Scenarist, for which I will always be grateful. Thank you DVD Maestro!

To answer the original question, having more frames in a GOP will reduce your encoding size. For example, if you were to compare an NTSC recording with 15 frames per GOP vs. the same video recorded at 10 frames per GOP, the one with 10 frames would be a lot larger. This is how the (in my opinion) bogus KVCD format is able to squeeze so much video into a short space. It uses crazy GOP sizes like 30-40 frames per GOP to achieve huge compression results.

[guns1inger]

Did you read my first post ? I said exactly that.

[Serville]

In that sense, the larger GOP size, the better compression.
So, for PAL GOP size 15 should be better than 12.
But I still don't see a reason why people say GOP 12 is optimal for PAL.

[adam]

I think those that suggest 12 just do so because its safer. With some encoders, setting GOP length to 15 could potentially cause some GOPs to be slightly too large. I've witnessed this with TMPGenc for example. If the encoder is doing its job than there's no reason not to use 15 in my opinion.

Realistically there's probably not going to be any noticable difference between using 12 or 15 anyway unless you are using a pretty low bitrate.

[gadgetguy]

I don't have an answer really, but I was intriqued by the question. Let me rephrase it the way I understood the question.

While I'm capturing video and it's being encoded on the fly, does GOP size make a difference?

My theory... I would think the smaller the GOP size, the less interpretation takes place allowing the encoder to keep up better while sacrificing file size. A larger GOP size would require more processing time to interpret the P and B frames possibly having the encoder not able to keep up resulting in lost data or dropped frames.

[SLK001]

I don't have an answer really, but I was intriqued by the question. Let me rephrase it the way I understood the question.

While I'm capturing video and it's being encoded on the fly, does GOP size make a difference?

My theory... I would think the smaller the GOP size, the less interpretation takes place allowing the encoder to keep up better while sacrificing file size. A larger GOP size would require more processing time to interpret the P and B frames possibly having the encoder not able to keep up resulting in lost data or dropped frames.

Just the opposite. The I frames require the most processing, followed by the P frames, then the B frames. If you make your GOP smaller, more I frames per second are produced, thus requiring more processing. P frames are less detailed, and B frames are even less detailed, so having more of these per second is less taxing on your processor.

[nmeuncn]

Excuse me, I am a little confused…I use adobe encore dvd 1.5 and I am interested only I quality of picture that is space is not a problem (1 hour per DVD is just enough). Any way it wants me to give him value for M frame and N frame. Default value is 3 for M frame and 12 for N frame. So If I got it right for the best video quality I should left 12 for N frame but I don’t know what value to choose for M frame. I haven’t mention that I use PAL standard. Please excuse me if I didn’t understand some of those you have already wrote above this text but I am a rooky and a little inexperienced and it is hard for me to understand everything…
Thank you,
nmeuncn

[themaster1]

I don't have an answer really, but I was intriqued by the question. Let me rephrase it the way I understood the question.

While I'm capturing video and it's being encoded on the fly, does GOP size make a difference?

My theory... I would think the smaller the GOP size, the less interpretation takes place allowing the encoder to keep up better while sacrificing file size. A larger GOP size would require more processing time to interpret the P and B frames possibly having the encoder not able to keep up resulting in lost data or dropped frames.

Yes it does when it's "on the fly" matter.
I use 1/1 values gop/sub gop size with windvr3 or fly 2000 tv for analog recordings coming from the cable/satellite and had great results.
it's also important to deinterlace (in my experience) rather than keep it interlaced still for analog cable/satellite recordings.
But of course this lead to a problem: the file is not dvd compliant.