JVET-Q0271 AHG9: On Subpicture Ordering [M. Damghanian, R. Sjöberg, M. Pettersson (Ericsson)]
This contribution was discussed Tuesday 7 January 2020 at 1800 (chaired by GJS).
This contribution proposes the following changes:
- To define an independent subpicture as a subpicture for which the subpicture boundaries are treated as picture boundaries in the decoding process and no loop filtering across the subpicture boundaries is done. (Editorial)
It was commented that this does not really seem necessary (esp. since aspect #3 is not adopted), but the question can be left to the editors for consideration.
- To add a syntax element sps_independent_subpics_flag in the SPS. When equal to 1 it specifies that all subpicture boundaries in the CLVS are treated as picture boundaries and there is no loop filtering across the subpicture boundaries. subpic_treated_as_pic_flag[ i ] and loop_filter_across_subpic_enabled_flag[ i ] are signalled only when sps_independent_subpics_flag is equal to 0. (Signalling efficiency)
This is just a syntax optimization, but it would be a shortcut for a common use case. It was commented that a common example for viewport-dependent streaming uses 96 subpictures, and this would avoid the need to send 2*96 flags.
Decision (cleanup): Adopt this aspect.
- To specify that the subpicture availability rule applies only when sps_independent_subpics_flag is equal to 0. (Functionality)
This change would allow arbitrary subpicture order in the bitstream when the subpictures are independent. Currently, subpicture order is constrained by the availability rule.
It was commented that this would impose a substantial buffering burden on decoder implementations with some architectures. Some decoders are envisioned to operate most of the decoding process in raster order. No action was taken on this.