GOP Sequence: What to use?

[sundance]

I tried two different GOP sequences in my PAL projects:
M=3, N/M=4 and M=3, N/M=5 (the first one comes from doom9's guide, the other one from another source I can't remember).

Is there a "correct" one?
I read that some players or authoring software are very picky when using the 3/5 sequence. But I also heard this concerning the empty AUDIO_TS folder, which (imho) cannot be necessary since there are a lot of commercial DVDs without such a folder. What's true then?

The reason why I would use the 5/3 sequence is that your movie takes up less disc space when you have less I-frames, or did I miss something here?

-sundance-

P.S. I use CCE 2.5 for encoding & DVDMaestro for authoring.

[slk001]

There is no correct one. I've heard of some problems with TMPG when encoding with 5/3, where the program will sometimes give you more in your GOP than 15 (I use CCE which uses this value as the MAXIMUM size of the GOP). 5/3will be a shorter file, while 4/3 will have slightly higher quality.

As for the AUDIO_TS folder, technically, it is part of the DVD specification and, for 100% complience, it HAS to be on the DVD. However, newer players don't necessarily need a 100% complient disk to be able to play the thing.

For the record, I use 4/3 for my GOPs almost exclusively (NTSC region), because 99% of all the DVDs that I've looked at are also 4/3.

[cjv]

I've personally had problems using 3/5, and never when using 3/4. I initially prefered to use 3/5 when I could, because you can fit more on the DVD-R without sacrificing noticible quality as much. (more P-frames, less I-frames)

I did a few Q 30 tests comparing 3/4 and 3/5. To my suprise, the difference was rarely > 20-30 megs, and the quality differences were non-existant.

Maestro will accept 3/5, but in my experiences, Scenarist will reject as an invalid file. You have nothing to lose, and the quality will be the same...I say go for 3/4 always.

cjv

[sundance]

slk001 & cjv,
thanks a lot for your info.
I was surprised that the gain in size was that small when using 3/5.
So, for me, I'd say there is no need to do anything else than 3/4.
slk001, btw., what tool did you use to tell the GOP sequence of a given mpeg stream?

-sundance-

[bira]

I only have problems with 3/5 if I encode as 23.976 and then use pulldown to get 29.97. If I encode directly to 29.97 I have no problems.

[slk001]

Originally posted by sundance
what tool did you use to tell the GOP sequence of a given mpeg stream?

I use Bitrate Viewer (full version - if you don't have it, you should get it).

I do not understand why someone would have problems with 3/5 and pulldown. Pulldown does NOT add any frames to a GOP. Someone commented that Scenarist would reject a 3/5 encode at 23.976 with the 3:2 pulldown. I suspect some bug in Scenarist if this is true (Maestro doesn't have this problem). I no longer use Scenarist, so I am really not interested in investigating this issue any longer.