Back to Search
Document details
CE2-related: Prediction dependent luma residual scaling for adaptive in-loop re-shaper
Abstract
The luma mapping with chroma scaling (LMCS) (previously referred to as adaptive in-loop re-shaper) was adopted into the VVC at Marrakech meeting. In the VVC working draft 5, the LMCS is applied to luma and chroma components in different ways: 1) the luma reshape is performed in sample domain using adaptive piecewise linear model; 2) chroma residuals are scaled based on the average of the corresponding luma prediction samples. To achieve one more unified design, this contribution proposes to extend the existing chroma residual scaling method to luma component in which the scaling factor of luma residuals are calculated based on the corresponding luma prediction samples. It is asserted that after applying the proposed method all the in-loop mapping of the predicted and reconstructed luma samples in the existing LMCS design can be removed. Compared to the VTM-5.0 anchors, the proposed method reportedly leads to BD-rate variations {0.04%, 0.06%, 0.12%} and {-0.05%, -0.12%, -0.13%} and {-0.06%, 0.01%, 0.02%} for AI ,RA and LD configurations, respectively.
JVET-O0603 CE2-related: Prediction dependent luma residual scaling for adaptive in-loop re-shaper [X. Xiu, Y.-W. Chen, T.-C. Ma, H.-J. Jhu, S. Ye, X. Wang (Kwai Inc.)]
Cited:
Decisions
JVET-O0603 CE2-related: Prediction dependent luma residual scaling for adaptive in-loop re-shaper [X. Xiu, Y.-W. Chen, T.-C. Ma, H.-J. Jhu, S. Ye, X. Wang (Kwai Inc.)]
Citation