JVET-Y0117 AHG12: ECM coding performance for HDR/WCG content and suggested common test conditions [T. Lu, F. Pu, P. Yin, S. McCarthy, W. Husak, T. Chen (Dolby), N. Hu, V. Seregin, M. Karczewicz (Qualcomm)]

This contribution presents HDR/WCG coding results for ECM3.1 with two encoder settings. The first setting follows the VTM HDR/WCG common test conditions. The second setting adds encoder configuration optimizations recently added for SDR coding with ECM and the CTU size of HLG sequences is 256. The results of ECM3.1 HDR coding compared with VTM11.0 with MCTF are as follows:

First setting (VTM HDR/WCG CTC):

For PQ content, the BD rate savings for wPSNRY/U/V/DE100/L100 are:

AI { -6.19% -18.43% -22.07% -12.31% -6.65% }

RA { -11.19% -25.44% -28.46% -19.21% -11.73% }

For HLG content, the BD rate savings for PSNRY/U/V are:

AI { -5.88% -12.24% -16.13% }

RA {-9.84% -18.70% -21.28% }

Second setting (VTM HDR/WCG CTC + ECM encoder optimization + CTU 256 for HLG):

For PQ content, the BD rate savings for wPSNRY/U/V/DE100/L100 are:

AI { -6.19% -18.43% -22.07% -12.31% -6.65% }

RA { -11.59% -27.82% -31.98% -22.98% -12.41% }

For HLG content, the BD rate savings for PSNRY/U/V are:

AI { -6.02% -11.86% -16.05% }

RA { -10.79% -20.84% -24.42% }

It was recommended to establish ECM HDR/WCG common test conditions corresponding to the second setting.

Decision (CTC): Include in CTC document for ECM/EE2, HDR as an optional class, and also change the VTM CTC for HDR (encoder optimization aligned with the ECM approach).

It was further suggested to seek for more HDR test material, in particular 4K for PQ would be desirable (e.g. content from VVC verification test).