[Mp4-tech] two questions on H.264 HRD

[Nancy Johnson]

Dr. Sullivan,
  Thanks for providing these valuable information. Regarding to the value of SchedSelIdx, 
  1. I believe the HRD is a golden model which will be used to test the conformance of bitstreams and real decoders.
  2. I believe the behavior of the HRD have been accurately specified in the spec due to 1. However, if I were asked to write a behavior model of the HRD by C or Verilog using the spec, I don't know how the HRD can determine the value of initial_cpb_removal_delay[SchedSelIdx] (at least it will influence the coded picture removal time from the CPB and further influence the decoded picture output time) because I can not find any information in the spec for the HRD to know the value of SchedSelIdx it should use.
  Maybe the HRD can derive an interpolated initial_cpb_removal_delay according to the actual bit rate it detected. Maybe the HRD is assumed to know an exact or interpolated value of SchedSelIdx from some system level info outside the bitstream. But this is not mentioned in the spec.
  I wish my question was invalid just because I misunderstood something or missed something.
  Thanks a lot.
  Best Regards,
  Nancy Johnson
Gary Sullivan <garysull windows.microsoft.com> wrote:
      Regarding question 1 -- I think you would need to read the HD DVD specs to find out, but personally I would not recommend trying to sell a lot of units of an HD DVD player that can't be output timing conforming at least the vast majority of the time for the vast majority of movie content.
  Regarding question 2 -- It is important to distinguish between the operation of a real decoder in a real system environment and decoder or bitstream unit testing for conformance.  For output timing conformance testing purposes, we assume that the HSS can force-feed the decoder in a manner consistent with the standard somehow, but we really don't care how.  It doesn't have to exclusively consist of the use of information within the bitstream.  Also note that HRD decoder testing is not necessarily performed by just selecting a value of SchedSelIdx -- there is also the possibility of "interpolated" scheduling.  There are really two different kinds of constraints specified in the standard -- checks that a conforming bitstream must be able to fulfill and checks that a conforming decoder must be able to fulfill.  It's important not to confuse the two.  Also, when operating in a real system environment, the system will ordinarily have some additional way of conveying the
 bitstream and associated timing information to the decoder.  In the case of HD DVD there are a significant amount of timing issues that are controlled at the MPEG-2 systems level rather than needing to rely on the information within the video elementary bitstream and I believe the system-level information and operation should ordinarily be considered more important to the actual operation of a real decoder in such an application environment.
  Best Regards,
  Gary Sullivan
---------------------------------
  From: mp4-tech-bounces lists.mpegif.org [mailto:mp4-tech-bounces lists.mpegif.org] On Behalf Of Nancy Johnson
Sent: Thursday, July 06, 2023 4:38 PM
To: mp4-tech lists.mpegif.org
Subject: [Mp4-tech] two questions on H.264 HRD
  Dear H.264 experts,
  Could you please help me to figure out these 2 questions?
  1. For HD-DVD decoding application, is it necessary for the H.264 decoder to claim output timing decoder conformance? It seems that for this case the decoder can be fed the bit stream in an "on demand" way.
  2. According to page 267 equation C-7 of H.264 03/2005, the removal time of the access unit from the CPB will be influenced by initial_cpb_removal_delay[SchedSelIdx]. Although initial_cpb_removal_delay[i] (i=0~cpb_cnt_minus1) are sent to HRD by buffering period SEI message (page 277 section D.1.1 of H.264 03/2005), I can not find a way in the H.264 stream definition for the HRD to know the specific value of SchedSelIdx in equation C-7. If this is true, how to make sure the value of SchedSelIdx in equation C-7 used by DUT are the same as that used by HRD? (Is this necessary for the decoder to meet output timing conformance?)
  Thanks a lot.
  Sincerely yours,
  Nancy Johnson
---------------------------------
  Sneak preview the all-new Yahoo.com. It's not radically different. Just radically better. 
---------------------------------
Want to be your own boss? Learn how on  Yahoo! Small Business. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/mp4-tech/attachments/20060707/3b059763/attachment-0001.html