Due to increasing resolutions and 360° capture, broadcast and video production is characterised by handling large data volumes. To ease such data intensive workflows a novel image and video codec called JPEG XS is currently standardised.
It focuses on film production, broadcast and ProAV markets and excels by ultra-low latency and ultra-low complexity. Moreover, it permits multiple encoding and decoding cycles with minimum quality loss and it can be implemented on different platforms such as CPU, GPU, FPGA and ASIC.
The paper describes these properties in more detail and explains how they help to devise optimal video and transmission workflows. Moreover, it details the used technologies in form of a block diagram and summarises the results of the quality evaluation ensuring visually lossless compression.
The goal of better visual experiences in the form of higher resolution and 360° movies causes transmission throughput in production networks to increase at a larger pace than the available network infrastructure.
This holds both for legacy infrastructures, whose replacement by a new generation is very costly, as well as for IP networks needing to simultaneously route multiple video streams. While standard video compression could be thought to solve these challenges, in practice existing standards such as JPEG, JPEG 2000 or HEVC do not comply with the needs of the film and broadcast production networks, as they are not designed for ultra-low latency and low complexity while achieving visually lossless compression.
Consequently their implementation costs are too high to justify their application. To overcome this situation, the JPEG Committee (formally known as ISO/IEC SC29 WG1) has been starting the standardisation of a novel compression codec called JPEG XS.