Opened 8 years ago

Last modified 7 years ago

#1475 new defect

Difference between HM and spec. on green metadata SEI

Reported by: Haruhiko Owned by:
Priority: minor Milestone: HM-16.15
Component: HM Version: HM-16.14
Keywords: Cc: ksuehring, davidf, karlsharman, jct-vc@…

Description

In the specification, green metadata SEI is not allowed in the SUFFIX_SEI_NUT.
But in the HM16.15, green metadata SEI write and read in the SUFFIX_SEI_NUT.

Change History (1)

comment:1 Changed 7 years ago by ksuehring

I agree, that the implementation does not match the HEVC spec.

I know very few about the use of the SEI because it is specified outside the HEVC spec.

But I'm curious about the current implementation: The scope of the SEI is "The CLVS containing the SEI message". So I guess, it would make sense, to include it at the beginning of the Coded Layer-wise Video Sequence (CLVS) to make the information available to the decoder when it is starting to decode the sequence.

But in the implementation we have:

    // Only produce the Green Metadata SEI message with the last picture.
    if( m_pcCfg->getSEIGreenMetadataInfoSEIEnable() && pcSlice->getPOC() == ( m_pcCfg->getFramesToBeEncoded() - 1 )  )

This tries to create the SEI only at the very last picture (which will fail in case IDR pictures are enabled, because POC is reset for IDR pictures).

It seems there is also no dependency on anything that is happening in the encoding process, which would require it to be written there.

So shouldn't it actually be included with any IRAP picture?

Note: See TracTickets for help on using tickets.

This list contains all users that will be notified about changes made to this ticket.

These roles will be notified: Reporter, Owner, Subscriber, Participant

  • David Flynn(Subscriber)
  • Haruhiko Tanaka(Reporter)
  • jct-vc@…(Subscriber)
  • Karl Sharman(Subscriber)
  • karl.sharman@…(Always)
  • Karsten Suehring(Subscriber, Participant, Always)