Fraunhofer IIS Explains JPEG XS For 8K Production

​JPEG XS is the new tailored solution for the production environment with specific requirements for highest quality, multi-generation robustness and low-latency. It’s open source, under ratification as a standard at the ITU and due by end of the year. Siegfried Foessel, Head of Department Moving Picture Technologies at Fraunhofer IIS, which submitted a codec in the call for proposals and proposed improvements during the standardization process, explains what the design of the new codec enables.

BroadcastBridge: Could you establish the context in which a codec like JPEG XS necessary?

Siegfried Foessel: Typically, image or video codecs are for a wide range of applications mainly targeting the distribution path. For professional production derivatives of these codecs with specific parameter settings are used but are not optimized for such use cases.
With JPEG XS the codec was specifically tailored for the use in professional production environments where highest quality is the main factor and not the compression ratio. In such applications typically a high bandwidth for transmission and storage is available, uncompressed transmission however is difficult, so a mezzanine compression fits the needs. JPEG XS is used in cases where 422 or 444 subsampling with up to 12 bit per component and compression ratio between 2:1 and 10:1 are necessary, low latency smaller than 32 lines and also multi-generation robustness is required. Typical example is the transmission for 4k60p which is roughly 10GBit/s. So to reduce the bandwith by a factor of 3 to 4 allows a safe transmission of multiple streams over a 10 Gbit Ethernet line.

What creative applications will it enable?

It is a cost effective solution to transmit HD60p over 1GBit Ethernet, multiple 4k60p streams over 10 Gbit Ethernet or 8k120p over 25Gbit Ethernet lines. In addition JPEG XS is optimized for different granularities of parallelism, so it can be used in an optimized way on FPGAs for embedding into hardware devices, for CPUs to work in editing workstations or on GPUs to monitor and display high resolution streams. For example we showed at the NAB a first prototype of an Adobe Premiere Plugin on a PC which can decode 4k60p JPEG XS files on a standard graphic card.

In general applications will be enabled where the bandwidth over wire is limited, real-time processing on standard PCs is necessary, or for multi-encoding and decoding processes in studio environments. In general transmission and processing of 4k or 8k resolutions with 60-120 frames per second will be enabled on standard PCs and low cost FPGAs.

What are the benchmark tests for JPEG XS performance?

On standard PCs we can reach even on the prototype a factor of 3-4 higher throughput than with JPEG2000 and we will expect even a higher performance after some more optimization steps. In any case 4k60p processing is possible on standard workstations. For low-latency applications (smaller 32 lines) the quality is comparable with JPEG 2000. In the case of rate allocation per frame (which is not in the JPEG XS standard right now, as this does not fulfill the low-latency requirement) it outperforms the quality of ProRes.

What is the timeline for its introduction to the market/ standardisation?

Part1 (The core coding system) is under DIS ballot until September. Then the technical aspects are frozen. Part 2 (Transmission and buffer models) are under CD ballot right now. Part-3 (Transport Layer and File formats) will be expected as CD in the July meeting. First implementation and introduction to the market will be expected by the end of 2018.

What is Fraunhofer's involvement in development?

JPEG XS is an open standard where multiple contributors share their knowledge to improve the quality of the standard. Fraunhofer submitted a codec in the call for proposals and proposed improvements during the standardization process. Fraunhofer is participating in the editing of the standards and shares software and improvements in the core experiments.

At NAB we showed a first Adobe Premiere plugin which demonstrated the real-time processing capabilities for 4K60p on a standard PC. In flicker tests the high quality of the codecs could be verified.

Is it compatible with SMPTE video over IP standards?

JPEG develops the coding core technology. Together with partner organization higher application layers will be developed. Under discussion are in the moment RTP payload over Internet, SDI encapsulation, MPEG-TS stream encapsulation and SMPTE 2110 packetizing. The codec is developed is such a way that it can be embedded in all of these higher layer protocols. The discussions are in different progress stages.

What is the essential technology change which drives improvements in compression?

For many new codecs and especially for JPEG XS the multi parallel processing capability is the driving factor. In the past many codecs had some sequential processing steps which made a parallel processing difficult or led to quality degradations. JPEG XS is developed to satisfy the parallel needs for multiple platforms. It can be easily implemented on FPGA using the multiple computation engines, it can be used on multi-core CPUs with a coarse or medium granularity and it can be used on GPUs with thousands of cores which needs a fine granularity. So processing power is in most cases not the limiting element it is more the access to it.

Let us know what you think…

Log-in or Register for free to post comments…

You might also like...

TV’s ‘Back to the Future’ Moment?

Philo T. Farnsworth was the original TV pioneer. When he transmitted the first picture from a camera to a receiver in another room in 1927, he exclaimed to technicians helping him, “There you are – electronic television!” What’s never been quoted but lik…

Remote Production: Synchronizing Video Sources Over The Internet

Live broadcasts are seen as nirvana in terms of attracting an audience. Presenting a live event, especially sports, in real-time and high quality, draws audiences like no other content. Yet, successfully originating these broadcasts is often both expensive and complex. ​A…

Cost-effective IP Contribution and Distribution

Saving dollars is one of the reasons broadcasters are moving to IP. Network speeds have now reached a level where real-time video and audio distribution is a realistic option. Taking this technology to another level, Rohde and Schwarz demonstrate in…

IMF - Interoperability and Content Exchange Made Easy

As the television business has become more global, and evolving consumer devices spawn the need for ever more formats, there has been an explosion of the number of versions that are needed for an item of content. The need to…

Myths and Truths About Standards Conversion

Everyone knows what standards converters do, right? Broadcast professionals recognize that changing the video format and frame rate is necessary when sharing materials internationally or when integrating movies into TV schedules. In fact, there are many types of standards conversion…