Question about consecutive BVOPs

[thop]

When you're analyzing a file for consecutive BVOPs is it safe to just analyze the first minute and assume that if lets say only 1 consecutive BVOP was detected then it is very unlikely that were will be 2 or more in other parts of the file (lets say a 120min file)? Same questions goes for packed bitstream, although i guess the answer is a clear yes here.

Sorry for that stupid question but i'm not really that familiar with the inner working of XViD and the AVI Container :)

[Aktan]

No, becuase xvid only adds b-frames as it sees fit. If in the first minute xvid doesn't decide to add more that 1 consec b-frames at a time, that does not mean later on in the video there will be a point where xvid decides 2 b-frames would be possible and add 2 b-frames. The same goes for packed bitstream. Of course xvid won't add more than the max allowed b-frame setting.

[akupenguin]

Packed bitstream is not adaptive (it has no effect on compression efficiency). It's either enabled or not, and that won't change throughout a movie. You can tell whether or not it's packed as soon as you see a B-frame.

(Actually, most options, including consecutive B-frames (and the presence of B-frames at all) can change at different points in the file, if you concatenate two videos encoded with different settings. But there's really nothing you can do about that case, and I would expect it to be pretty rare. OTOH, I have encoded some such movies myself, e.g. where qpel helped on the credits but hurt compression of the movie itself.)

[Koepi]

The CC bframe amount can change _any_ time.

Just imagine you encode with max bframes set to 15. So most of the time max 1 or 2 CC bframes will be used. But then this one scene close before the end of the movie where nothing happens suddenly the amount of cc bframes raises to i.e. 5.

So you have to do a whole scan on the file to know for sure.

Regards
Koepi

[thop]

I thought so, unfortunately time is critical so i'd rather only scan a short part. Could someone give me a good time to scan to be more or less on the safe side? I'm only looking for cc BVOPs > 1, if it's 2 or 10 doesn't really matter anymore then.

I've tried a few movies and so far 5 minutes seems to be enough to catch 2 cc BVOPs.

[akupenguin]

I would expect you could get greater certainty per amount of computation by scanning short intervals chosen randomly throughout the file, rather than just the first N minutes.

[Aktan]

Originally posted by akupenguin
Packed bitstream is not adaptive (it has no effect on compression efficiency). It's either enabled or not, and that won't change throughout a movie. You can tell whether or not it's packed as soon as you see a B-frame.


When I said "same" goes for packed bitstream, I meant B-frames are inserted the same with packed bitstream enabled or not. I did not mean packed bitstream is adaptive. Sorry for the confusion.