FW: [Mp4-tech] standardization of h.264 hardware motionestimationoutput

[MPEGIF List Admins]

Another bounced mail. 
Alex, please change your address from dolby.net to dolby.com if this is your sending address going forward.  (Go to   <http://lists.mpegif.org/mailman/listinfo/mp4-tech> http://lists.mpegif.org/mailman/listinfo/mp4-tech and enter old email all the way down)
  _____  
From: Tourapis, Alexis [mailto:alexis.tourapis dolby.com] 
Sent: Friday, 15 December 2023 11:25
To: Mario Tapilouw; mp4-tech lists.mpegif.org
Subject: RE: [Mp4-tech] standardization of h.264 hardware motionestimationoutput
Dear Mario,
Motion estimation is not a normative process. Your motion estimation module can generate what it believes is best for your requirements (i.e., quality, speed, complexity, power etc). For example, a module may select to only perform and generate 16x16 motion vectors. This could penalize coding efficiency but would still be valid. A different module could spit out all possible information for all modes, or only a "final" best determined by a mode decision process undertaken within the module. Note that this information (modes and motion vector values) can be passed to other modules in any form you wish (i.e., in actual numbers). The transformation to semantics and entropy coding as specified in the standard for coding modes and motion vectors in reality needs to be performed at the very last stage, i.e. within your syntax/entropy coding module and you do not need to "code" this information. Note that there are at times benefits in knowing the number of bits that may be required for coding such information (i.e. for performing RD optimal coding decisions), but nothing is forcing you to (even in such a case) output information in such a way. 
Best regards,
Alexis
PS. I would suggest checking the reference encoder and the encode_one_macroblock function to see how this process is done in that software. 
  _____  
From: mp4-tech-bounces lists.mpegif.org [mailto:mp4-tech-bounces lists.mpegif.org] On Behalf Of Mario Tapilouw
Sent: Thursday, December 14, 2023 7:06 PM
To: mp4-tech lists.mpegif.org
Subject: [Mp4-tech] standardization of h.264 hardware motion estimationoutput
Dear All, 
I am developing a hardware motion estimation for H.264. I am looking for a reference that could help me form the output of my hardware, i.e. block type and it's appropriate motion vectors. Are there any rules for standardized output format of hardware variable block size motion estimation? especially on how a macroblock is coded (16x16, 2 16x8, 2 8x16, 4 8x8, etc. ). 
I have read the ITU-T H.264, there are mb_type stated on table Table 7‑13 and sub_mb_type[ mbPartIdx ] on table 7-17. I wonder if these are what I've been looking for?  
Thank you very much.
mario 
  _____  
Need a quick answer? Get one in minutes from people who know. Ask your question on Yahoo!  <http://answers.yahoo.com/;_ylc=X3oDMTFvbGNhMGE3BF9TAzM5NjU0NTEwOARfcwMzOTY1NDUxMDMEc2VjA21haWxfdGFnbGluZQRzbGsDbWFpbF90YWcx> Answers. 
  _____  
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/20061215/db5017e5/attachment.html