JVET-C0102 Luma delta QP adjustment based on video statistical information [J. Kim, J. Lee, E. Alshina, Y. Park (Samsung)] [late]

This contribution is identical to JCTVC-W0039. The contribution is targeting performance improvement of high dynamic range video compression using existing HEVC tools. It was asserted that the performance can be improved by adjusting the luma QP based on video statistics. Depending on the average of the luma components in a picture and the luma block average and variance, the algorithm adjusts the delta QP for each CTU. The performance of the algorithm is measured with the same method of SuperAnchor v3.2 and reportedly shows consistently better compression performance.

One expert pointed out that with this method enabled, it might be even more difficult to use objective metrics.

Further study in an AHG on JEM coding of extended colour volume material was requested.

Motion compensation and vector coding (13)

JVET-D0165 Cross-check of JVET-D0098 Decoupling decoder-side intra mode derivation from entropy decoding [Alexey Filippov, Vasily Rufitskiy, Haitao Yang (Huawei)] [late]

EE9: Adaptive Scaling for Extended Colour Volume (2)

Primary (2)

From summary report JVET-D0010:

This experiment studies an alternative to the residual scaling proposed in JVET-B0054. The motivation in JVET-B0054 is to remove the overhead of delta QP that is signalled by the encoder to compensate for the reshaping of quantization noise as a function of brightness in next generation containers such as ST 2084.

Questions recommended to be answered during EE tests:

[Q]: What is performance of JVET-C0066, JVET-C0095 and JVET-C0102 using the content and test conditions provided by the AhG 7 on JEM coding of extended colour material with:

• Results reported with the weighted BD-Rate and BD-PSNR metrics, where the weighting corresponds to the target luma-to-QP mapping function.
• Results reported with BD-Rate and BD-PSNR for the tPNSRY and deltaE metrics, where the metrics are computed from linear RGB
• Results reported for visual quality assessment

[A]: Only partial data available.

[Q]: What is the benefit of combining different aspects of the JVET-C0066, JVET-C0095 and JVET-C0102?

[A]: Not answered yet.

Conclusion from EE summary: The test procedure should be agreed first. AhG7 was discussing testing procedure.

BoG (coordinated by A. Segall) recommendation: Review EE results, and agree conversion method and test procedure

See further discussion under JVET-D0192: Both proposals are to be further investigated in new EE6.

JVET-C1011 Description of Exploration Experiments on coding tools [E. Alshina, J. Boyce, Y. J. Chang, S.-H. Kim, V. Seregin, X. Xiu] [2016-06-23]

See list of EEs in section 9.1

The version reviewed on Tuesday was again approved in the closing plenary on Wednesday. One minor change was made, adding Qualcomm as an additional contributing party in EE9 test 4.

Some affine MC friendly content should be made available, as allowed by content rights, so that it can be used in EE4 tests. X. Ma (Huawei) was requested to contact Mathias Wien to enable uploading of such content to the JVET content repository. Once content has been made available, this should be announced on the JVET reflector.