Opened 5 years ago

Closed 4 years ago

#746 closed defect (fixed)

Issue with tc derivation on deblocking filter

Reported by: abe.kiyo Owned by:
Priority: minor Milestone:
Component: spec Version: VVC D7 vE
Keywords: Cc: ksuehring, bbross, XiangLi, fbossen, jvet@…

Description

In section 8.8.3.6.1, the variable tc for luma is derived by the equation (1265):
tc = BitDepth < 10 ? ( tc′ + 2 ) >> ( 10 − BitDepth ) : tc′ * ( 1 << ( BitDepth − 10 ) )

In section 8.8.3.6.3, the variable tc for chroma is derived by the equation (1325):
tc = ( BitDepth < 10 ) ? ( tc′ + 2 ) >> ( 10 − BitDepth ) : tc′ * ( 1 << ( BitDepth − 8 ) )

''8'' should be ''10''. VTM is already implemented that way.
''2'' should be ''1 << ( 10 - BitDepth − 1 )''. VTM needs to be modified as well.

Change history (3)

comment:1 Changed 5 years ago by Kenneth

Thanks for spotting the issue in equation 1325. I agree that 8 shall be changed to 10.

Regarding "2" that is a rounding offset to achieve identical tc when BitDepth is 8. So that part is correct.

comment:2 Changed 5 years ago by abe.kiyo

Thank you for the comment.
Regarding "2", I agree that this issue is only for other than 8bit. So it is not necessary.
However, similar processing in other tools are described like as follows.
shift = 10 − BitDepth
offset = 1 << ( shift − 1 )
I think it would be a little more appropriate to align this description as well.

Last edited 5 years ago by abe.kiyo (previous) (diff)

comment:3 Changed 4 years ago by bbross

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

Thanks for reporting.

This has been fixed in JVET-Q2001-v8.

Note: See TracTickets for help on using tickets.