JVET-R0239 AHG9: Cleanups on signalling of tiles, slices, and subpictures [K. Zhang, L. Zhang, Y.-K. Wang, Z. Deng, J. Xu, H. Liu (Bytedance)]
Aspect 1 is no longer relevant because of the action taken for JVET-R0062. For aspects 2 and 5, see the the notes for JVET-R0415.
Aspects 3 and 4 were discussed on 24 April at 0900 UTC (chaired by GJS).
Aspect 3: When num_exp_tile_columns_minus1 is equal to PicWidthInCtbsY − 1, tile_column_width_minus1[ i ] is not signalled and inferred to be 0. Likewise, when num_exp_tile_rows_minus1 is equal to PicHeightInCtbsY − 1, tile_row_height_minus1[ i ] is not signalled and inferred to be 0. This aspect is for an unusual “corner case”, and it was agreed that no action is needed for it.
Aspect 4: The semantis of exp_slice_height_in_ctus_minus1[ i ] is clarified that exp_slice_height_in_ctus_minus1[ i ][ num_exp_slices_in_tile[ i ] − 1 ] is used to derive the uniform slice height, instead of specifying the height of the (num_exp_slices_in_tile[ i ] − 1) -th slice.
Decision (editorial bug fix / expression of existing intent): Adopt aspect 4.