Mismatch between SW and Spec on setting of half sample interpolation filter index in HMVP table

[LiZhang]

In the SW, the HMVP table is updated with the half sample interpolation filter index associated with a block, and the index of a HMVP candidate could be inherited by another block. However, the first part is missing in the current draft.

Suggested draft changes are as follows:
1) In clause 8.5.2.1, adding the filter index as an input parameter:
The updating process for the history-based motion vector predictor list as specified in clause 8.5.2.16 is invoked with luma motion vectors mvL0[ 0 ][ 0 ] and mvL1[ 0 ][ 0 ], reference indices refIdxL0 and refIdxL1, prediction list utilization flags predFlagL0[ 0 ][ 0 ] and predFlagL1[ 0 ][ 0 ], half sample interpolation filter index hpelIfIdx, and bi-prediction weight index bcwIdx.

2) add the filter index as an input parameter in clause 8.5.2.16 and set a HMVP candidate being associated with the index.

8.5.2.16 Updating process for the history-based motion vector predictor candidate list
Inputs to this process are:
– luma motion vectors in 1/16 fractional-sample accuracy mvL0 and mvL1,
– reference indices refIdxL0 and refIdxL1,
– prediction list utilization flags predFlagL0 and predFlagL1,
– half sample interpolation filter index hpelIfIdx,
– bi-prediction weight index bcwIdx.

The MVP candidate hMvpCand consists of the luma motion vectors mvL0 and mvL1, the reference indices refIdxL0 and refIdxL1, the prediction list utilization flags predFlagL0 and predFlagL1, the half sample interpolation filter index hpelIfIdx, and the bi-prediction weight index bcwIdx.

[vdrugeon]

Ticket #595 has been marked as a duplicate of this ticket.

[bbross]

Thanks for reporting.

This will be fixed with the missed integration of P0856 in JVET-P2001-vA.