Software and Hardware Encoding of Omnidirectional 8K Video: A Performance Study

Software and Hardware Encoding of Omnidirectional 8K Video: A Performance Study

Stať ve sborníku v databázi WoS či Scopus

Omnidirectional (360-degree) video streaming has been on the rise dramatically in recent years. As a result, demand on the monitoring and evaluation of the visual quality of such videos is rapidly increasing. This paper deals with the performance study of hardware and software encoders for 360° video in resolution 8K, especially in terms of encoding speed on a consumer computer. New encoding standards significantly improve encoding efficiency, but also increase encoding time due to increased computational complexity. In this work, different implementations of the High Efficiency Video Codec (HEVC) for 360° videos at 8K resolution are compared. Next, quality presets that balance visual quality and encoding speed at the same bitrate are also considered. Netflix's Video Multimethod Assessment Fusion (VMAF) objective metric is used to measure video quality. Furthermore, attention is devoted to the study of power consumption and encoding speed, considering CPU and GPU performance limitations. The results show that hardware encoders can encode 8K 360° videos in real-time with quality comparable to software encoding using x265 compression. However, the encoding speed is approximately 20 times faster.

Omnidirectional (360-degree) video streaming has been on the rise dramatically in recent years. As a result, demand on the monitoring and evaluation of the visual quality of such videos is rapidly increasing. This paper deals with the performance study of hardware and software encoders for 360° video in resolution 8K, especially in terms of encoding speed on a consumer computer. New encoding standards significantly improve encoding efficiency, but also increase encoding time due to increased computational complexity. In this work, different implementations of the High Efficiency Video Codec (HEVC) for 360° videos at 8K resolution are compared. Next, quality presets that balance visual quality and encoding speed at the same bitrate are also considered. Netflix's Video Multimethod Assessment Fusion (VMAF) objective metric is used to measure video quality. Furthermore, attention is devoted to the study of power consumption and encoding speed, considering CPU and GPU performance limitations. The results show that hardware encoders can encode 8K 360° videos in real-time with quality comparable to software encoding using x265 compression. However, the encoding speed is approximately 20 times faster.

8K; virtual reality; 360° video; objective video quality, VMAF

8K; virtual reality; 360° video; objective video quality, VMAF

KUFA, J.; POLÁK, L.; ŠIMKA, M.; ŠTECH, A.

2024

19.04.2023

Pardubice

979-8-3503-9834-2

33rd International Conference Radioelektronika

1

5

5

https://ieeexplore.ieee.org/document/10109035

http://hdl.handle.net/

2023078024