[Mp4-tech] question about "complentary filed pair" a nd "field_pic_flag"

[Gary Sullivan]

Zhou Xiao,
Regarding question 1, I kind of like the word amphibolous.  It seems like a combination of amphibian and bulbous, so I envision a rather lumpy and bloated frog.  That may be an interesting analogy for field handling in the standard - many people do not understand this part the first time they read it.  The concept of field pairing exists in the standard primarily as a way to deal with memory structuring and memory referencing in a decoder.  When we conceptually pair two fields together in the decoding process, the pair of fields can be referenced for the decoding of a subsequent frame.
Regarding question 2, I don't know exactly what you mean by one field being "of one frame" and another field being "of the following frame".  It seems to me that the definition of a field being "of a frame" would correspond to the field having a particular value of frame_num.  But fields are not considered complementary if they have different values of frame_num.  So perhaps the answer to your question is No.  But I am not sure that I am confident that I understand what you are asking.
Regarding the third (not numbered) question, field_pic_flag must be the same in all slices of a picture.  There are several such things in the slice header that must be the same in all slices of the picture.  If you think the slice header is not where we should have put that flag, where do you think would have been better?  This is part of the design of the "parameter set" concept of the standard.  The idea is that the things that can change frequently from picture to picture are placed in the slice header for reasons of robustness to data losses.  If the video is being sent on a channel that does not have many data losses, then the encoder can choose to send just one slice for the entire picture -- so the idea of the slice header resulting in a waste of bits is not really relevant with such use.  Overall, I think we did a very good job of figuring out where to put the various syntax elements.  And I think the notions of the sequence and picture parameter sets are very good ones.
Overall, I would say that the field and frame concepts in the standard may take a significant amount of studying before you fully understand them. Don't expect to understand everying just by reading some parts of the standard just once and guessing at what is meant.  The text of the standard (if you get the latest version) is very clear about most things, but people often feel that they must read it over and over before they really understand it completely.
Note that most overview presentations about the standard (e.g., books and articles) do not really cover such details as.
Best Regards,
Gary Sullivan
________________________________
From: mp4-tech-bounces lists.mpegif.org on behalf of ZhouXiao
Sent: Sat 6/10/2023 9:04 AM
To: mp4-tech lists.mpegif.org
Subject: [Mp4-tech] question about "complentary filed pair" a nd "field_pic_flag"
Hi all,
I am still confused about the concept of "complentary filed pair".
(1) Why such a concept exists? Its defination and application seem very amphibolous, Why we have to introduce such a concept? that is: in which condition will a complementary ref field pair or a complementary non-ref field pair occur, and if this condition is true, what special operation will the decoder take? 
(2) Is it possible that the bottom field of one frame and the top field of the following frame immediately in decoding order constitute a complementary ref field pair or a complementary non-ref field pair?
Another question is about field_pic_flag.
As you know, the field_pic_flag is just defined in the slice header layer, 
now supposing frame_mbs_only_flag == 0 && mb_adaptive_frame_field_flag == 0 ( i.e.: using PAFF), is it possible that one slice in a frame adopts field coding mode, while the other slice in the same frame adpotes frame coding mode? 
Maybe it sounds very stupid. and the possible answer is NO. But if that, it seems inappropriate that the field_pic_flag syntax element is put in the slice layer.
? ? ? ? ? ? ? ? "?" ? ? ? ? ? 
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?,? ? Ajax ? ?,126 "D ? ?"? ? ? ? ? ! <http://www.126.com/> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/mp4-tech/attachments/20060611/25d1074e/attachment.html