Truncation issue on subpicture signaling

[geonjungko]

This ticket reports an issue on semantics of subpic_ctu_top_left_x, subpic_ctu_top_left_y, subpic_width_minus1 and subpic_height_minus1.
The length of above-mentioned syntax elements and the inferred value are determined based on pic_width_max_in_luma_samples, pic_height_max_in_luma_samples and CtbSizeY.

According to the subclause “5.2 Arithmetic operators”, “/” results in the truncated integer value. Therefore, if the picture width is not a multiple of CtbSizeY, ( pic_width_max_in_luma_samples / CtbSizeY ) cannot represent the rightmost CTB. It is the same for vertical direction.

Suggested fix: change “/” to “÷” as below

subpic_ctu_top_left_x[ i ] specifies horizontal position of top left CTU of i-th subpicture in unit of CtbSizeY. The length of the syntax element is Ceil( Log2( pic_width_max_in_luma_samples ÷ CtbSizeY ) ) bits. When not present, the value of subpic_ctu_top_left_x[ i ] is inferred to be equal to 0.

subpic_ctu_top_left_y[ i ] specifies vertical position of top left CTU of i-th subpicture in unit of CtbSizeY. The length of the syntax element is Ceil( Log2( pic_height_max_in_luma_samples ÷ CtbSizeY ) ) bits. When not present, the value of subpic_ctu_top_left_y[ i ] is inferred to be equal to 0.

subpic_width_minus1[ i ] plus 1 specifies the width of the i-th subpicture in units of CtbSizeY. The length of the syntax element is Ceil( Log2( pic_width_max_in_luma_samples ÷ CtbSizeY ) ) bits. When not present, the value of subpic_width_minus1[ i ] is inferred to be equal to Ceil( pic_width_max_in_luma_samples ÷ CtbSizeY ) − 1.

subpic_height_minus1[ i ] plus 1 specifies the height of the i-th subpicture in units of CtbSizeY. The length of the syntax element is Ceil( Log2( pic_height_max_in_luma_samples ÷ CtbSizeY ) ) bits. When not present, the value of subpic_height_minus1[ i ] is inferred to be equal to Ceil( pic_height_max_in_luma_samples ÷ CtbSizeY ) − 1.

[vdrugeon]

Solved by adoption of JVET-Q0787