JVET-Y0107 AHG9: Text improvement for the film grain SEI [E. Thomas (Xiaomi)]
It is asserted that the description text and formula of the film grain SEI message may be improved for the reader of the specification. The text is written with the assumption of a 4:4:4 coded sequence while there is no normative constrain of that nature when using this SEI. Hence, using equations with variables such as PicHeightInLumaSamples and c for the different components tends to hide the prior assumption made at the beginning of the clause.
A possible text modification is proposed to make more explicit the 4:4:4 convention used throughout the equations.
Revision 1: After offline discussion, it was clarified that there was no bug in the specification text but that the readability of the text may be improved.
Decision (ed.): Adopt JVET-Y0107 for V2 of VSEI.
Generally it was agreed that editorial changes would help understanding the SEI message.
In the discussion, it is however agreed to keep Pic<Width|Height>InLumaSamples unchanged, as this is commonly used in VVC/HEVC. Further, I-prime should better become I-hat to denote the image that is input to the process. That image should not be called “decoded image” but better “input image” (which may have been converted from the decoded picture).
The above intent was to be implemented by editors.