Small issues and typos in affine and subblock TMVP specification text

[vdrugeon]

In the semantics of five_minus_max_num_subblock_merge_cand, the value of MaxNumMergeCand is overwritten. I believe that the equation should calculate MaxNumSubblockMergeCand instead of MaxNumMergeCand.

Figure 7 1 – Multi-type tree spliting modes indicated by MttSplitMode (informative) seems to have been removed, but the caption is still in the document. Was it intentional?

In section 8.3.3.2 Derivation process for luma affine control point motion vectors and reference indices in affine subblock merge mode, when both sps_sbtmvp_enabled_flag and sps_affine_enabled_flag are true for the sequence, numSbX and numSbY are set to cbWidth >> 2 and cbHeight >> 2 respectively in step 2, therefore overwritting the values calculated for these two variables in step 1. Since the values for numSbX and numSbY are different for SBTMVP and affine, I think that there should be two sets of numSbX and numSbY variables: one for SBTMVP and one for affine. Alternatively, the values of numSbX and numSbY could be reset at the end of the section in case the selected merge candidate is equal to SbCol.

In section 8.3.3.6 Derivation process for constructed affine control point motion vector merging candidates, when deriving the fourth (collocated bottom-right) control point motion vector cpMvLXCorner[ 3 ], the derivation process for collocated motion vectors as specified in clause 8.3.2.10 is invoked with refIdxLX as input. I cannot find where this variable has been defined. Should it be refIdxLXCorner[ 3 ] instead of refIdxLX?

[bbross]

Fixed in v4 of VVC D4