JVET-AN0190 AHG17/AHG10: On test model simulating hardware encoder [Y. Zhao, J. Mao, R. Zhao, X. Ma, T. Solovyev, E. Alshina (Huawei)]

See section 4.16.1.

JVET-AN0190 AHG17/AHG10: On test model simulating hardware encoder [Y. Zhao, J. Mao, R. Zhao, X. Ma, T. Solovyev, E. Alshina (Huawei)]

Hardware encoders have become increasingly important for video applications, and they are deployed on billions of mobile devices with constrained chip area size and power consumption. It is recommended to develop the next-generation video coding standard of a native design for hardware-friendly encoding. A test model developed with an encoding configuration that mimics some key characteristics of hardware encoder C model (besides a configuration for simulating software encoder) may facilitate more accurate evaluation on the performance of coding tools in the context of hardware encoder implementation and solicit hardware encoding friendly coding technologies.

Several experts supported the idea of putting more focus on hardware-friendly implementation. It was however commented that investigation of hardware complexity of encoders is more complicated than for decoders, as the degrees of freedom are much higher.

Examples might be limitation on number of RDO decisions needed, granularity of partitioning, memory access limitations.

Smartphones (or mobile devices in general) are mentioned as important. Also hardware transcoding is used a lot by content/service providers.

JVET-AN0118 presents a software module for counting RDO decision in ECM.

It was concluded to install a new AHG on hardware implementation complexity (Y. Zhao, L. Li, X. Li, …)

CfE submissions and evaluation (11)

Contributions in this area were discussed during 0910–1330 and during 1415–1845 on Tuesday 7 Oct. 2025 (chaired by JRO).