[Mp4-tech] Regarding MBAFF flag

Daniel, Ilan Ilan.Daniel lsi.com
Thu May 22 16:04:08 EDT 2008


Following Alexis good explanation of MBAFF functionality,
I just wanted to add that using MBAFF suppose to give better compression
for interlace material,
therefore clair.cif is not a good representative for testing MBAFF
performance since it is not interlaced in its nature.
Regards,
Ilan
________________________________
From: mp4-tech-bounces lists.mpegif.org
[mailto:mp4-tech-bounces lists.mpegif.org] On Behalf Of Tourapis, Alexis
Sent: Thursday, May 22, 2024 3:32 AM
To: aditi G; mp4-tech lists.mpegif.org
Subject: RE: [Mp4-tech] Regarding MBAFF flag
Dear Aditi,
I am assuming you are talking about MPEG-4 AVC/H.264 and not H.263 given
your email. In any case, if MBAFF is enabled
(mb_adaptive_frame_field_flag set to 1 in the sequence parameter sets
and field_pic_flag set to 0 in the slice header syntax) does not dictate
that a macroblock, or in fact a pair of macroblocks in this particular
scenario, will be coded in field mode. Instead you have the option to
select if this pair of macroblocks is to be coded in either field or
frame mode. This decision could be based on a variety of methods, like
for example the one used in the JM which involves encoding the two
macroblocks (which in this scenario basically constitute an area of 16 x
32 pixels) in field and frame mode and then selecting the option that
gives the best rate distortion performance. Methods exist where this
16x32 area is analyzed, i.e., with certain filters, and given this
analysis a decision is performed on what mode to use to encode these
macroblocks. There are also several other methods which perform partial
encoding, i.e., examine motion characteristics (assuming we are talking
about inter modes), and base the encoding again on such analysis and so
on.
I hope this helps you in your analysis of MBAFF.
Best regards,
Alexis
________________________________
From: mp4-tech-bounces lists.mpegif.org
[mailto:mp4-tech-bounces lists.mpegif.org] On Behalf Of aditi G
Sent: Wednesday, May 21, 2024 9:26 AM
To: mp4-tech lists.mpegif.org
Subject: [Mp4-tech] Regarding MBAFF flag
Dear Experts,
I have some doubts regarding MBAFF implementation in H.263.
If MBAFF is enabled, I know that macro blocks will be arranged in field
format rather than frame format.
I need to know that, the input video file, like clair.cif, need to be in
field scanned format?
How the macro blocks are arranged, while I am calculating the nC value
in CAVLC?
I read the H.264 spec and trying to understand the JM reference related
to this topic. But I didn't get enough.
If any one has the theory /concept behind MBAFF, I request to forward
that to me.
Regards,
AD
________________________________
This message (including any attachments) may contain confidential
information intended for a specific individual and purpose. If you are
not the intended recipient, delete this message. If you are not the
intended recipient, disclosing, copying, distributing, or taking any
action based on this message is strictly prohibited. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/mp4-tech/attachments/20080522/5229c402/attachment.html


More information about the Mp4-tech mailing list