Back to Search Document details
39th Meeting: Daejeon, KR, March 2025 2025-04-03 20:31
AHG9: Colour mapping information SEI
Abstract
A colour mapping information SEI message is proposed to enable description of the use of colour maps that transform monochrome samples into RGB samples that have been selected for improved human visualization. Related contribution JVET-AL0218 proposes adding a colour map codepoint to CICP. The SEI message proposed in this contribution signals a colour map ID that is interpreted by the CICP colour map codepoint.
JVET-AL0219 AHG9: Colour mapping information SEI [J. Boyce, T. Biatek, M. M. Hannuksela (Nokia)]

Discussed on 3 April 2025 at 17:50 chaired by S. Deshpande.

A colour mapping information SEI message is proposed to enable description of the use of colour maps that transform monochrome samples into RGB samples that have been selected for improved human visualization. Related contribution JVET-AL0218 proposes adding a colour map codepoint to CICP. The SEI message proposed in this contribution signals a colour map ID that is interpreted by the CICP colour map codepoint.

This contribution also supports use of colour map IDs that are “Unspecified” by CICP and allows their colour map function to be determined by external means or to be described in the SEI message via a piecewise linear look up table.

The proposed syntax currently assumes something is added to CICP, but it can be modified to only provide lookup if CICP addition was not done.

It was asked why SEI needs to specify which colour map to apply and can the client not choose which colour map it wants to apply. It was answered that the content creator intent can be signalled via the SEI.

It was suggested to use EOI type for this purpose. It was answered that the mapping information does not exist in EOI, but it could be presumably added.

It was commented that there are two modes and one of those may apply only to monochrome/ luma.

It was commented that the luma min-max might be different based on if it is full range or not. It was asked why min value needs to be signalled as it is available in video range. It was done for easy mapping purpose.

It was commented that if we decide to do this two separate SEIs should be considered.

It was agreed to treat the decision about SEI separately than CICP decision.

The proponent suggested to add this SEI to TuC and to further study CICP aspect.

It was suggested to remove cmi_map_for_display_flag. It was ok to keep that flag for now and in next meeting we should consider if that flag is needed or should be removed.

It was agreed to remove the CICP dependency, the semantics of cmi_colour_map_id will be changed and conditioning on value > 192 will be removed

A revision of the text will be uploaded to reflect this.

Decision: Agreed to include in TuC with changes described above to be uploaded in a revision.

Decisions
Agreed to include in TuC with changes described above to be uploaded in a revision
Citation