JVET-O0247 AHG17: Multiple ALF APS for chroma [V. Seregin, N. Hu, M. Coban, M. Karczewicz (Qualcomm)]
In the current VVC, only one ALF APS can be used for chroma in a slice, while luma component may use multiple APS and an APS index is signalled in CTB to indicate the utilized filter coefficients. This contribution proposes to enable the same functionality for chroma ALF, i.e. to signal APS Ids for chroma in a slice header and to signal a chroma APS id per CTB when ALF is applied. The second aspect of the contribution is to remove redundant APS index signalling in a CTU when only 2 APS are used.
It is agreed that switching between several chroma filters is desirable (results of CE5 indicate that up to 8 are beneficial)
Following options:
- Stay with current APS (luma set + 1 chroma), but allow several APS in slice and select at CTU which one is best for luma and which one is best for chroma
- Extend current APS such that several chroma filters are in, but if at CTU level different luma filter sets shall be selected, it may mean that chroma filters need to be sent in several APS
- Have separate APS for chroma, either 1 per filter, or 1 with multiple filters
The most flexible solution would be to have an APS that contains 0…25 luma filters and 0..8(or more?) for chroma. This would allow most flexibility in usage, including the case where an APS just carries a collection of chroma filters.
See further discussion in Monday plenary
JVET-O0246 is also related but proposes separate ALF APS for luma and chroma. It is noted that this might provide more flexibility in inheriting filters from other pictures, but would also make the signalling for selection more complicated.
If only 1 chroma filter is in the APS, would it be a disadvantage to send a set of chroma filters in separate NALU?
The second aspect of the contribution suggests to save the signalling of the APS index in cases where the number is <= 2. This is a minor cleanup. Decision: Adopt this aspect.