JVET-AJ0321 AHG12: ECM complexity trade-offs [Y.-J. Chang, C.-C. Chen, M. Coban, K. Cui, P. Garus, N. Hu, H. Huang, M. Karczewicz, J.-L. Lin, P.-H. Lin, P. Nikitin, B. Ray, V. Seregin, G. Verba, H. Wang, E. Ye, R. Yu, Y. Zhang, Z. Zhang (Qualcomm)] [late]

This contribution presents two sets of the results (a) ECM-14.0 without inter template matching, simplifications of other mainly decoder side ECM tools, and encoder optimizations; and (b) encoder only optimizations for ECM-14.0 keeping all tools enabled and unmodified.

Test results comparing to VTM-11.0ecm14.0 anchor are summarized as follows.

Test (a) RA: -20.78%, -20.43%, -23.25%; 412.4%, 606.1% (YUV, encT, decT)

Test (b) RA: -24.72%, -24.83%, -27.70%; 466.9%, 1022.3% (YUV, encT, decT)

This was for information only – no request was made by proponents to add the software to ECM.

High-level syntax (HLS) and related proposals (66)

AHG9: Aspects on SEI messages in VSEIv4 and related (32)

General comments (5)

Contributions in this area were discussed at 1500–1730 on Friday 1 Nov. 2024 (chaired by J. Boyce).