JVET-AL0070 AHG9: On packed regions information SEI message [T. Chujoh, Z. Fan, R. Ishimoto, T. Ikai (Sharp)]

1. The loop's variable i has been started from 1. An index equal to 0 is not used.
2. The syntax element pri_resampling_ratio_idx[ i ] in the array is misspelled.
3. In the syntax table of pri_multilayer_flag, pri_region_layer_id[ i ] and pri_region_is_a_layer_flag[ i ] are related. But, in the semantics of pri_multilayer_flag, pri_region_is_a_layer_flag is not described.
4. When pri_num_resampling_ratios_minus1 equals 0, the syntax element pri_resampling_ratio_idx[ i ] does not exist; however, the inferred value is not defined.
5. When pri_target_pic_params_present_flag equals 0, the syntax elements pri_target_pic_width_minus1 and pri_target_pic_height_minus1 do not exist; however, the inferred values are not defined.
6. When pri_targe_pic_params_present_flag equals 0, the syntax elements pri_target_region_top_left_x[i] and pri_target_region_top_left_y[i] do not exist. However, the target picture cannot be reconstructed without those syntax elements.
7. The regions in the input picture are packed, not overlapped. It is believed that the target regions should be defined.
8. The values of pri_target_pic_width_minus1 and pri_target_pic_height_minus1 should be constrained depending on the chroma format.
9. Depending on the chroma format, the values of the top-left sample position, the width, and the height of the rectangle regions should be constrained. However, those specifications are incomplete.
10. When pri_multilayer_flag equals 1, multiple layers exist. However, the basic semantics of PRI SEI seem to consider a single layer.

Regarding Item 1, the loop index is intentionally started from 1 with values for index 0 inferred.

JVET-AL0129 is also related. Further discussion after JVET-AL0129 is reviewed.

Item 2 agreed.

Item 3 agreed. JVET-AL0132 Item 2 also proposes.

Regarding Item 4, it was questioned if inference is needed, because the syntax element is not used. JVET-AL0122 Item 2 also proposes. Further discussion after checking if an inference value is needed.

Regarding Items 5 and 6, it was suggested that it isn’t always necessary to reconstruct a target picture, in which case reconstruction of the target picture isn’t needed. No action on items 5 and 6.

Item 7 is related to JVET-AL0129 item 3 and JVET-AL0122 Item 11. A bug has been identified in that the position in the target picture rather than the coded picture should be used. Should look at all proposals to select how to fix. Further discussion.

Item 8 is related to JVET-AL0122 item 10 and JVET-AL0098 item 4. If no action is taken there, this item should be agreed. Further discussion.

Item 9 proposes 3 solution options to enforcing chroma alignment. JVET-AL0122 item 7 relates to the second part of solution 1. JVET-AL0129 item 11 is related to the first part of solution 1.

Y.-K. Wang to provide a summary of related PRI proposals to make it easier to compare the alternative related proposals.