[Mp4-tech][H.264] Sliding window decoded reference picture marking

[yaguang xie]

Hi Gary,
  Thanks very much for your clear interpretation!
  But I am still confused for something and have another question.
  Is the following case possible?
  There are 6 successive reference fields in the bitstream as: 
  field 1(top,frame_num=1),field 2(bottom,frame_num=1),
  field 3(top,frame_num=2),field 4(bottom,frame_num=3),
  field 5(top,frame_num=4),field 6(bottom,frame_num=4).
  If it is possible, field 3 and field 4 are different frame, and the
  both frames will have only one field? It seems be conflicted with 
  defenition 3.53(A frame consists of 2 field,top field and bottom field).
  and also I think it is not clear that how to output these "one field frame" to 
  PC screen.(maybe output to TV is OK).
  If the case is impossible, I cannot find which section could guarantee
  the impossibility in the standard.
  Best regards,
  Yaguang Xie
Gary Sullivan <garysull windows.microsoft.com> wrote:
      Yaguang Xie,
  It is not entirely clear what you mean by "two fields of a frame".  I believe that the only way this concept appears in the decoding process in the standard is in the concept of complementary field pairs.  But both fields of a complementary field pair have the same value of frame_num, while you are referring to "two fields of a frame ... with different frame_num", so your question does not seem to make obvious sense.
  In any case I think the basic answer to your question is that if there are two reference fields in the DPB and those two fields have different values of frame_num, then those fields cannot be considered to be a complementary reference field pair.  They would therefore, according to your description, seem to be categorized as two non-paired reference fields.  And they would therefore count as 2 rather than 1 in the computation of numShortTerm.
  This interpretation should be particularly clear if you read Annex C.  The concept is that the decoder has frame storage areas.  When two fields are classified as a complementary field pair, they both go into the same frame storage area.  When they are not complementary (such as when they have different values of frame_num), they go into different frame storage areas.  The sliding window process manages these frame storage areas as its primitive units for reference picture marking, so it counts two of them for two non-paired reference fields and it counts one of them for a complementary reference field pair.
  Best Regards,
  Gary Sullivan
---------------------------------
  From: mp4-tech-bounces lists.mpegif.org [mailto:mp4-tech-bounces lists.mpegif.org] On Behalf Of yaguang xie
Sent: Thursday, June 22, 2023 9:00 PM
To: mp4-tech lists.mpegif.org
Subject: [Mp4-tech][H.264] Sliding window decoded reference picture marking
  Dear experts,
  I have a question to 8.2.5.3(sliding window ref_pic marking) as following.
  Let numShortTerm be the total number frames,complementary reference field pairs and non-paired reference fields for which at least one field is marked as "used for short-term reference".
  My question is: If the two fields of a frame are short-term reference picture with different frame_num,  should the numShortTerm be added by 1 or 2 for this frame?
  (since the frame_num is different, the two fields should not be complementary,they are non-paired reference fields.)
  Thanks in advance!
  yaguang xie
---------------------------------
  Yahoo! Sports Fantasy Football ¡¯06 - Go with the leader. Start your league today! 
---------------------------------
How low will we go? Check out Yahoo! Messenger’s low  PC-to-Phone call rates.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/mp4-tech/attachments/20060624/80255eba/attachment-0001.html