Opened 5 years ago
Closed 5 years ago
#463 closed defect (fixed)
incorrect size of p[ x ] and mWeight[ x ][ y ] for MIP in 8.4.5.2.1
Reported by: | lixinwei | Owned by: | |
---|---|---|---|
Priority: | minor | Milestone: | |
Component: | spec | Version: | VVC D6 vE |
Keywords: | Cc: | ksuehring, XiangLi, fbossen, jvet@…, bbross, jvet@… |
Description
8.4.5.2.1 Matrix-based intra sample prediction
current specification in JVET-O2001-vE:
For the generation of the input samples p[ x ] with x = 0..2 * inSize − 1, the following applies
shall be corrected to:
For the generation of the input samples p[ x ] with x = 0..inSize − 1, the following applies
and current specification in JVET-O2001-vE:
– The weight matrix mWeight[ x ][ y ] with x = 0..2 * inSize − 1, y = 0..predC * predC − 1 is derived by invoking the MIP weight matrix derivation process as specified in clause 8.4.5.2.3 with MipSizeId[ xTbCmp ][ yTbCmp ] and modeId as inputs.
shall be corrected to:
– The weight matrix mWeight[ x ][ y ] with x = 0..inSize − 1, y = 0..predC * predC − 1 is derived by invoking the MIP weight matrix derivation process as specified in clause 8.4.5.2.3 with MipSizeId[ xTbCmp ][ yTbCmp ] and modeId as inputs.
Because the variable inSize is the size of the input samples p[x], so no need to multiply 2.
Change history (1)
comment:1 Changed 5 years ago by bbross
- Resolution set to fixed
- Status changed from new to closed
Good catch! Thanks!
Fixed in a document we are preparing to be submitted as an editorial input
to the 16th JVET meeting.