Back to Search Document details
16th Meeting: Geneva, October 2019 2019-10-02 18:49
[AHG15] On scaling list for JCCR
Abstract
In the scaling process for transform coefficients in the current VVC draft specification, chroma QP offset Qp′CbCr is introduced to apply to the TU coded with JCCR. It is because the residual energy distribution of the transform block with JCCR might be different from that without JCCR. Nevertheless, the current draft specification does not support scaling list for the transform block coded with JCCR. This contribution proposes scaling list for transform blocks with joint coding of chroma residual (JCCR). When a given transform block is coded with JCCR and both tu_cbf_cb and tu_cbf_cr are equal to 1, scaling list for JCCR are used for the scaling process. An alternate solution is also proposed to avoid signalling of additional scaling lists. The solution consists in deriving the scaling factors in case of JCCR by averaging the scaling factors applying to the Cb and Cr components.
JVET-P0608 [AHG15] On scaling list for JCCR [S. Iwamura, S. Nemoto, A. Ichigaya (NHK), P. de Lagrange, F. Le Leannec, E. François (InterDigital)]

In the scaling process for transform coefficients in the current VVC draft specification, chroma QP offset Qp′CbCr is introduced to apply to the TU coded with JCCR. It is because the residual energy distribution of the transform block with JCCR might be different from that without JCCR. Nevertheless, the current draft specification does not support scaling list for the transform block coded with JCCR. This contribution proposes scaling list for transform blocks with joint coding of chroma residual (JCCR). When a given transform block is coded with JCCR and both tu_cbf_cb and tu_cbf_cr are equal to 1, scaling list for JCCR are used for the scaling process. An alternate solution is also proposed to avoid signalling of additional scaling lists. The solution consists in deriving the scaling factors in case of JCCR by averaging the scaling factors applying to the Cb and Cr components.

Not evident that there is really need to have a separate scaling list for mode 2 of JCbCr. For mode 2, there is some indication that both components are similar anyway.

This would require additional generating and storing the average of the two scaling lists which every decoder would need to implement, whereas only few encoders might use scaling lists and optimize wrt. JCbCr.

No action was taken on this.

PATENTS:
US20220217339A1 0.34 2022-07-07 US12149694B2 0.32 2024-11-19 CN114556924B 0.32 2024-01-26 CN114402601B 0.30 2024-02-27 EP4068782B1 0.28 2025-05-21 CN113810690B 0.26 2022-04-29 CN114556955B 0.24 2024-04-05 CN114503588B 0.22 2024-10-08
Decisions
No action was taken on this.
Citation