JVET-AK0208 AHG9: On the inclusion of quality metric SEI message in SEI processing order SEI message [H. Tan, C. Kim, J. Nam, J. Lee, J. Lim, S. Kim (LGE), M. M. Hannuksela (Nokia)]

This contribution proposes the several changes to the semantics of quality metric (QM) SEI message. It is asserted that the proposed changes provide improvement to solve the asserted problems related to the inclusion of QM SEI message in SEI processing order (SPO) SEI message and when it is wrapped as PON-nested SEI message.

The proposed changes are as follows:

1. Clarify that QM SEI message may be present in the bitstream as non-PON-nested SEI message, PON-nested SEI message, or both.
2. Specify that PON-nested QM SEI messages that are part of the same processing chain (i.e., included in a particular sei processing order SEI message) have different scope / domain from the PON-nested QM SEI message that are part of another processing chain.
3. Clarify the association between QM SEI message that is included in an SPO SEI message to its presence in PON SEI message based on its processing order and prefix information.
4. Add a constraint such that when a QM SEI message is included in an SPO SEI message, the QM SEI message shall not be the first SEI message included in the SPO SEI message.
5. Update the current text such that the quality metric information that is signalled in the QM SEI message that is the i–th SEI message in an SPO SEI message indicates picture quality resulting from post-processing after invocation of all other SEI messages in the SPO SEI message that have po_sei_processing_order[ j ] equal to or lower than the value of po_sei_processing_order[ i ].

Regarding Item 1, it is noted that the current design already allows the QM SEI to be present in a PON and/or not nested in a PON. It seems unnecessary to add the proposed wording.

Item 2 is a clarification. There was agreement to Item 2, with some potential editing to refine the wording.

It was suggested that the language proposed in Item 3 is already handled in the language for the SPO SEI.

Item 4 is withdrawn.

Item 5 is seen to already be the design intent and may already be addressed in JVET-AK0205.

It was noted that the original design intent was to be able to signal quality differences associated with processing of SEI messages. Care would need to be taken to avoid affecting the design intent.

Discussed again for item 5 follow-up on 19 January 2025 at 19:05 PM chaired by S. Deshpande.

After offline discussion v2 is available and was discussed.

Definition from general section of interface text needs to be moved for handling the proposed text correctly. This is delegated to the editors.

The text for item 5 in v2 was agreed.

Decision: Adopt Item 2, Item 5 (from v2).

Other (10)

Contributions in this area were discussed at 1010–1300 on Saturday 18 Jan. 2025 (chaired by J. Boyce) and 1900 to 2000 on Sunday 19 Jan. 2025.