JVET-K0445 Compression performance report of VTM/BMS for 8K test sequences [S. . Iwamura, S. . Nemoto, A. . Ichigaya (NHK)]
This contribution reports on compression performance of BMS-1.1 associated with 8K test sequences. The clips span a wide range of spatial temporal characteristics and portray content typical of broadcasting programs. Three ITU-R BT.2020 SDR test sequences have been encoded with HEVC test model (HM-16.16) and VVC test model (BMS-1.1 with VTM/BMS configurations). The simulation results show similar tendency of bit rate saving and encoding/decoding complexity for 8K test sequences to those for CTC sequences. It is confirmed that current BMS-1.1 software works for 8K sequences as expected.
This contribution was pPresented in track B Tuesday 17th 10:00 (chaired by JRO).
The 8K test sequences used for this test are parts of the test materials “Ultra-high definition/wide-color-gamut standard test sequences – Series B”, distributed by the Japanese academic society, the Institute of Image Information and Television Engineers (ITE). Note Tthat these 8K test sequences were only provided to the licensee and not allowed to be redistributed. They therefore cannot be used by JVET for standard development.
Results with RA configuration (only 33 frames, i.e. 2 IRAPs were encoded at the beginning and end, which approximately corresponds to 1/4 s RA period. The gGain compared to the HM is reportedly somewhat less than what we currently see for 4K, which may however be due to the fact that more intra coding is used.
BMS-1.1 with VTM config over HM-16.16
BMS-1.1 with BMS config over HM-16.16
This was interesting information – the contributors said they will conduct this test from time to time to report on how future versions VTM/BMS perform on 8K material.