Opened 5 years ago
Closed 4 years ago
#938 closed defect (fixed)
HMVP candidate indexing bug in 8.5.2.6 and 8.6.2.4
| Reported by: | hangao | Owned by: | |
|---|---|---|---|
| Priority: | minor | Milestone: | |
| Component: | spec | Version: | VVC D8 vB |
| Keywords: | Cc: | ksuehring, bbross, XiangLi, fbossen, jvet@… |
Description
In JVET-Q2001-vD
In 8.5.2.6 and 8.6.2.4 the for loop of HMVP candidate list usage:
For each candidate in HmvpCandList[ hMvpIdx ] with index hMvpIdx = 1..NumHmvpCand, the following ordered steps are repeated until numCurrMergeCand is equal to MaxNumMergeCand − 1:
However the candidate is scanned from last to first, and HmvpCandList[ NumHmvpCand ] is not exist.
Suggest fix:
For each candidate in HmvpCandList[ NumHmvpIbcCand − hMvpIdx ] with index hMvpIdx = 1..NumHmvpCand, the following ordered steps are repeated until numCurrMergeCand is equal to MaxNumMergeCand − 1:
Change history (1)
comment:1 Changed 4 years ago by bbross
- Resolution set to fixed
- Status changed from new to closed
Note: See TracTickets for help on using tickets.
Good catch, thanks!
This will be fixed in first verison of D10 but with different fixes for both occurances (IBC and non-iBC).