Issue with 6.4.2 Allowed quad split process

[shih-ta.hsiang]

Two problems with the allowed quad split process:
1.When a CTU block exceeds the right or bottom picture boundary and both the CTU size and min QT size are set equal to 128, it is inferred to be further split while having all split types disabled.
2.In the allowed quad split process, cbSize/SubWidthC is used for size comparison against MinQtSizeC, which is measured in luma samples.

In Marrakech’s meeting, inferred QT split to avoid 32x128/128x32 partitions at picture boundaries was adopted (JVET-M0446/M0888/M0905). Problem 1 is not present in the adopted method (as shown in the proposed spec text in JVET-M0888) but was introduced during the spec text integration process.

The proposed fix is as follows:
Add one input to the process
– a location ( x0, y0 ) of the top-left luma sample of the considered coding block relative to the top-left luma sample of the picture,

The variable allowSplitQt is derived as follows:
– If one or more of the following conditions are true, allowSplitQt is set equal to TRUE:
– x0 + cbSize is greater than pic_width_in_luma_samples and cbSize is greater than 64
– y0 + cbSize is greater than pic_height_in_luma_samples and cbSize is greater than 64
– Otherwise, if one or more of the following conditions are true, allowSplitQt is set equal to FALSE:
– treeType is equal to SINGLE_TREE or DUAL_TREE_LUMA and cbSize is less than or equal to MinQtSizeY
– treeType is equal to DUAL_TREE_CHROMA and cbSize is less than or equal to MinQtSizeC
– mttDepth is not equal to 0
– treeType is equal to DUAL_TREE_CHROMA and ( cbSize / SubWidthC ) is less than or equal to 4
– treeType is equal to DUAL_TREE_CHROMA and modeType is equal to MODE_TYPE_INTRA
– Otherwise, allowSplitQt is set equal to TRUE.

The input variables in 7.4.10.4 Coding tree semantics should be updated accordingly.

[XiangLi]

The issue is resolved in VVC D8 (JVET-Q2001-v3) with the adoption of JVET-Q0330.