Better to Have No B-Pictures?

[aamir12345678]

Is it better to have no B-pictures at all? Does this result in better quality? Can anyone explain why? If a P-picture is the same size as a B-picture, which one yields better overall quality for the video?

[petar]

Is it better to have no B-pictures at all?

Generally no.

Does this result in better quality?

On the contrary.

Can anyone explain why?

B pictures provide the best compression. Though one may think a less compressed picture gives better quality, if your bitrate was not limited that would be true. But you only have so much bitrate to play with.

[adam]

You are confusing B pictures with P pictures. P's have the greatest compression because they only look to the previous reference frame, whereas B's look either forward, backward or both.

B pictures are like instant gratification. Like I said, they reference more frames than P pictures, therefore they provide a more accurate representation of the source for that particular frame. But B pictures cannot be a reference frame, only I and P's can. So when you use a B picture that's it, it does not contribute anything to the quality of your encode beyond that single frame. P frames, on the contrary, can be referenced by following P pictures or by preceding or following B pictures. So the more P pictures you use, the better quality your other P and B pictures will be. Does this mean you shouldn't use B pictures? Only if you have very low bitrate then sometimes eliminating B pictures will be beneficial. Otherwise, its good to have some, and at very high bitrates its good to have more but only if your decoder can handle it. B pictures have alot more overhead and are much more complex to decode.

For more information check out the sections titled, "what do B-pictures buy you?" and "Why do some people hate B-pictures."
http://bmrc.berkeley.edu/frame/research/mpeg/mpeg2faq.html

[petar]

No, use of B pictures provides the highest compression for all the reasons you mentioned (with which I agree 100%). If a frame can be compressed using both forward and backward information, it can achieve higher compression than one that only uses backward information, no? The fact that that frame can no longer be used as reference is irrelevant to the fact that it compressed better. However using too many B frames in a row will obviously affect the quality because the next P frame has to reference a frame that's "too far" from it. But eliminating B frames for this reason (to improve the P picture quality) is not going to help the quality of the overall video. Because B pictures compress better (by better I mean for the same compression they give better quality) they use less bitrate and the bitrate is the overall constraint on quality.

Hope now I make more sense.

[adam]

Uh oh, you're right

B's carry much more overhead (computationally) but physically take up less space, aka require less bitrate. I don't know why I got that confused. I like the description that the Berkley site gives... "digital spackle."

One thing we definitely agree on is that you shouldn't completely eliminate B pictures for quality's sake. But for the sake of the decoder there are reasons. Lots of commercial broadcasts have no B pictures because of the increased overhead. I sometimes forgo B pictures on some of my projects which are to be played back by media player classic (I love it but it crashes constantly...using no B pictures seems to help and requires less CPU power for older pc's which some of my target audience uses.)

[aamir12345678]

Thanks guys.

[petar]

But for the sake of the decoder there are reasons.

True.


Taken from www.mpeg.org

"The user can arrange the picture types in a video sequence with a high degree of flexibility to suit diverse applications requirements. As a general rule, a video sequence coded using I-pictures only (I I I I I I .....) allows the highest degree of random access, FF/FR and editability, but achieves only low compression. A sequence coded with a regular I-picture update and no B-pictures (i.e I P P P P P P I P P P P ...) achieves moderate compression and a certain degree of random access and FF/FR functionality. Incorporation of all three pictures types, as i.e. depicted in Figure 8 (I B B P B B P B B I B B P ...), may achieve high compression and reasonable random access and FF/FR functionality but also increases the coding delay significantly. This delay may not be tolerable for e.g. videotelephony or videoconferencing applications."