JVET-R0263 AHG9: On TemporalId and sublayer [Y. He, V. Seregin, M. Coban, M. Karczewicz (Qualcomm)]

This contribution was discussed on 23 April at 1440 (chaired by GJS).

This contribution proposes the constraints on TemporalId as follows:

1. TemporalId value is in ascending order for sublayers of the same CLVS. It was remarked that the n-th sublayer is the layer with TemporalId equal to n, and that this is already clear in the text.
2. The value of TemporalId shall be the same for all VCL NAL units of a PU or an AU if mixed_nalu_types_in_pic_flag is equal to 0; otherwise, the value of TemporalId may be different. This would relax a constraint for mixed NAL unit type. It was commented that this is not consistent with the design intent and would have a substantial impact on the text (e.g., extraction should be possible based on the TemporalId).
3. The value of TemporalId shall be the same within each sublayer of a CLVS if there are multiple sublayers, the value of TemporalId may be different if there is a single sublayer. It was commented that this affects the definition of what is a sublayer; a sublayer is the set of NAL units with a particular value of TemporalId.
4. Sublayer definition update. This was also proposed as a change for the mixed case. Again it was not considered consistent with the design intent.
5. Additional modification options are proposed keep the PTL structure consistent as follows.

• Option1: signal maximum number of sublayers in VPS and pass it to VPS PTL structure
• Option2: signal maximum TemporalId in SPS and pass it to SPS PTL structure

The motivation for most of these items is to require less coordination of sub-bitstream encoding for merging use cases. However, it was commented that coordination is expected as the design intent and these proposed changes would be substantial and likely to introduce problems (for example in the sub-bitstream extraction process). For the last item, participants felt that the requested functionality of gaps in presence of sublayers is already embodied in the text and did not see a clear problem. So no action was taken on this.

Profile, tier, level (PTL) (5)

Discussion started here for AHG Session 1.5 on Tuesday 7 April at 1300 (chaired by GJS, YKW, JRO & JB).