[Mp4-tech] Reference Lists

[Gary Sullivan]

K.J. Nitthilan et al,
See response in-line below:
+> -----Original Message-----
+> From: mp4-tech-bounces lists.mpegif.org 
+> [mailto:mp4-tech-bounces lists.mpegif.org] On Behalf Of 
+> Nitthilan Kannappan Jayakodi
+> Sent: Tuesday, March 01, 2024 1:27 AM
+> To: mp4-tech lists.mpegif.org
+> Subject: [Mp4-tech] Reference Lists
+> 
+> Dear experts,
+>       In the h.264 standard, two reference lists(0 and 1) 
+> are used. Why
+>       is this so? because,
+> 1) the reference pictures are present in both the lists
They can be present in both lists, but they do not necessarily need to
be.  That is up to the discretion of an encoder to decide.
(Also note that there is little meaningful difference between being at
the end of a list without being referred to by reference index and not
being in the list at all.)
+> 2) the forward , backward and bi-prediction can be done from a single
+> list
Hypothetically, that might be true in some design.  Whether some other
design along those lines would be as good as the design we actually have
in this standard is different question.  More detail about the
alternative design would need to be provided and studied.
+> 
+> further if u say that the two reference lists the ordering 
+> is such that
+> the neighbouring pictures( the immediate fw. and backward 
+> picture) may
+> have a less reference index ( mostly zero) and thus produce 
+> compression,
+> why to repeat the pictures in the two lists.
The encoder determines what is in the final lists.  If it wants to
remove some picture(s) from some list(s), it can.  Also, if the pictures
at the end of a list are never used, it basically makes approximately no
functional difference whether they are in the list or not.
However, sometimes using multiple references into the same reference
picture can provide a coding efficiency benefit.
See, for example, the papers and the book published by Markus Flierl,
Thomas Wiegand, and/or Bern Girod on the subject of "multi-hypothesis"
motion compensation.
Additionally, even if the encoder is not using the same picture for both
references, there may be other reasons for wanting some picture to be in
both lists.
For example, if each picture is only allowed to be in one list and there
are two pictures that you want to use for bi-prediction, but you have
placed both of them in the same list, then you would not be able to do
what you want.
There is also weighted prediction to consider. 
And when a list contains more than two reference pictures, there is no
loss of coding efficiency involved in making the list longer.
And there is no significant increase in complexity that is required to
support the idea of letting pictures be in both lists.
+>    We can have List(1) for all backward pictures and List(0) 
+> for all fw.
+>    pictures.
That sounds like it would result in just old MPEG-2-style operation.
Allowing more than two reference pictures has a coding efficiency
advantage.
Note that the uses of the words "backward" and "forward" are not really
necessarily applicable anymore.
Best Regards,
Gary Sullivan
+> 
+> 
+> Is there any special reason for this kind of arrangement? 
+> Pls. clarify.
+> 
+> expecting ur reply,
+> K.J.Nitthilan
+> -- 
+>   Nitthilan Kannappan Jayakodi
+>   nitthilan fastmail.fm
+> 
+> _______________________________________________
+> NOTE: Please use clear subject lines for your posts. Include 
+> [audio, [video], [systems], [general] or another 
+> apppropriate identifier to indicate the type of question you have.
+> 
+> Note: Conduct on the mailing list is subject to the 
+> Antitrust guidelines found at 
+> http://www.mpegif.org/public/documents/vault/mp-out-30042-Ant
+> itrust.php
+>