Opened 5 years ago

Closed 4 years ago

#1035 closed defect (fixed)

RPL constraint for RADL picture

Reported by: hendry197 Owned by:
Priority: minor Milestone: VVC D10
Component: spec Version: VVC D9 vB
Keywords: Cc: ksuehring, bbross, XiangLi, fbossen, jvet@…

Description

The following constraint is currently in the spec.

When the current picture is a RADL picture, there shall be no active entry in RefPicList[ 0 ] or RefPicList[ 1 ] that is any of the following:

  • A RASL picture
  • A picture that precedes the associated IRAP picture in decoding order

Currently a RASL PU is defined as picture that contain slices that are all RASL_NUT or mixed of RASL_NUT and RADL_NUT. Based on this, I think the constraint above should be updated to be as follows to be more precise:

When the current picture is a RADL picture, there shall be no active entry in RefPicList[ 0 ] or RefPicList[ 1 ] that is any of the following:

  • A RASL picture with mixed_nalu_types_in_pic_flag is equal to 0
  • A picture that precedes the associated IRAP picture in decoding order

Change history (6)

comment:1 Changed 5 years ago by yk

According to the suggestion, it means that an active entry of the RPLs of a RADL picture can refer to a RASL picture with mixed_nalu_types_in_pic_flag is equal to 1. Should that be allowed? If that is allowed, then starting decoding from the associated CRA picture the RADL picture may or may not be fully correct. That's does not seem a correct behavior.

comment:2 Changed 5 years ago by hendry197

When decoding starts from the associated CRA picture, RASL picture would still be decoded. If the RASL pic has mixed_nalu_types_in_pic_flag is equal to 1, the RASL slice(s) in that picture will not be corrected decoded since they refer to unavailable ref pics. However, the RADL slice(s) in that picture will be correctly decoded since they should not refer to anything that is not available. We have constraint that say RADL subpicture cannot refer to RASL subpicture or any picture that precedes the CRA in decoding order.

comment:3 Changed 5 years ago by yk

But still we need to be careful of the wording, as the following wording can be interpreted that the slices for which the collocated slices in the reference RASL picture (with mixed_nalu_types_in_pic_flag equal to 1) that are RASL slices can also have the reference picture in its active RPL entries (which should not be allowed):

When the current picture is a RADL picture, there shall be no active entry in RefPicList[ 0 ] or RefPicList[ 1 ] that is any of the following:
A RASL picture with mixed_nalu_types_in_pic_flag is equal to 0
A picture that precedes the associated IRAP picture in decoding order

We should think a bit more on this, and probably address it through a contribution to the AHG meeting in three weeks.

comment:4 Changed 5 years ago by hendry197

agreed. we can think a bit of better wording for this.

comment:5 Changed 4 years ago by bbross

  • Milestone set to VVC D10
  • Version set to VVC D9 vB

comment:6 Changed 4 years ago by yk

  • Resolution set to fixed
  • Status changed from new to closed

Resolved at the May HLS AHG meeting.

Note: See TracTickets for help on using tickets.