Opened 5 years ago
Closed 5 years ago
#307 closed defect (fixed)
DMVR subblock boundary is processed as subblock boudnary in deblocking filter
| Reported by: | chhuanb | Owned by: | |
|---|---|---|---|
| Priority: | minor | Milestone: | VVC D5 v8 |
| Component: | spec | Version: | VVC D5 v7 |
| Keywords: | Cc: | ksuehring, bbross, XiangLi, fbossen, jvet@… |
Description
In "clause 8.5.1, step 5", NumSbX[ xCb ][ yCb ] and NumSbY[ xCb ][ yCb ] are set equal to numSbX and numSbY, which are used in clause "8.8.3.4 Derivation process of coding subblock boundary".
If the coding block applies DMVR, numSbX and numSbY might be greater than 1, and then clause 8.8.3.4 is applied for that coding block.
While in VTM5, only the subblock boundary of SbTMVP and affine coded block is filtered.
Change history (3)
comment:1 Changed 5 years ago by Kenneth
comment:2 Changed 5 years ago by bbross
- Milestone set to VVC D5 v8
comment:3 Changed 5 years ago by bbross
- Resolution set to fixed
- Status changed from new to closed
Fixed in D5 v8
Note: See TracTickets for help on using tickets.
Thanks for reporting.
One approach to fix this is to check if inter_affine_flag or merge_subblock_flag are equal to 1 and only then set numSbX=NumSbX[xCb][yCb] and numSbY=NumSbY[xCb][yCb], otherwise set numSbX and numSbY to 1.