Performance Study of AV1 Encoders and Presets for 360-degree Video in Virtual Reality

Kufa, J.; Polak, L.; Kratochvil, T.; Ieee,

Recently, due to the pandemic of coronavirus (known as COVID-19), travel options for people are very limited. However, technology Virtual Reality (VR), one of the emerging fields in the multimedia services, can help to overcome these barriers. Providing of a VR-based video content in a form of online streaming has very high requirements on the network capacity. Hence, advanced video compression algorithms with high efficiency for delivery of a video in the format of VR arc vital. In this paper, we present an objective-based performance study of different implementations of the AOMedia Video 1 (AV1) video codec for 360-degree videos with a resolution of UHD. Different quality presets that set the balance between the visual quality and encoding speed for the same bitrate arc also considered. The Video Multimethod Assessment Fusion (VMAF) objective metric, developed by Netflix, is used to measure the quality of the video sequences. Results show that all the tested implementations of the AV1 encoder have a problem with the real-time encoding of the 360-degree videos.