Opened 6 months ago

Closed 2 months ago

#415 closed defect (fixed)

Chroma QP offset in loop filter

Reported by: adarsh Owned by:
Priority: minor Milestone:
Component: spec Version:
Keywords: Cc: ksuehring, bbross, XiangLi, fbossen, jvet@…

Description

Under discussion of JVET-O0650, it was agreed to apply the chroma offsets after the mapping table. This was implemented in QP derivation for chroma, but missed in the loop filter QP derivation.

Current Spec.:
qPi = Clip3( 0, 63, ( ( QpQ + QpP + 1 ) >> 1 ) + cQpPicOffset ) (8-1132)
QpC = ChromaQpTable[ cIdx − 1 ][ qPi ] (8-1133)

It should be:
qpi = ChromaQpTable[ cIdx − 1 ][ ( QpQ + QpP + 1 ) >> 1 ) ] (8-1132)
qPi = Clip3( 0, 63, qpi + cQpPicOffset ) (8-1133)

Change history (2)

comment:1 Changed 3 months ago by bbross

Is this still an issue after adoption of JVET-P1001?
"Non-CE5: Chroma QP derivation fix for deblocking filter (Combination of JVET-P0105 and JVET-P0539)"

comment:2 Changed 2 months ago by bbross

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

Confirmed by Adarsh, ticket can be closed.

Note: See TracTickets for help on using tickets.