Compression is an essential enabler for video delivery over real-world networks and JPEG XS provides new options for premium, high-bitrate contribution (including ‘Ground to Cloud’ remote production) and studio distribution applications.
Although it may be tempting to consider JPEG XS a generic standard simply because the video codec underpinning it is generic, the reality could not be further from the truth: Any deployable solution will contain layers of functionality beyond the encode / decode elements which are critical and ripe for differentiation. These aspects have a critical role in defining overall system capability, even in pivotal aspects such end-to-end latency. This paper aims to highlight the most important of these with the aim of demonstrating that not all JPEG XS solutions are the same!
The fundamental objective of JPEG XS
Today’s broadcast industry is highly differentiated which is why a range of codecs exist to provide best fit to any application. When contributing premium sports and events content from an event to a studio, or to the cloud, broadcasters are sometimes able to achieve the technical ideal and transmit the content uncompressed.
Uncompressed is synonymous with ‘uncompromised’, meaning transmission takes place without the need for complex encoding / decoding equipment and, even more importantly, having to bear typical compression drawbacks such as decreased VQ and increased latency. For these reasons, the ability to omit an encoding / decoding stage and send material uncompressed is the reference standard for the industry, but even though networks continue to advance offering higher bandwidth at less cost, it is becoming harder to contemplate sending content uncompressed because the number of feeds needing to be contributed is increasing dramatically alongside equally dramatic increases in video resolution.
The uncompressed bitrate of modern video standards is staggering: The 270Mb/s uncompressed rate of SD has largely been superseded by HD 1080i which demands an uncompressed bitrate increase of approximately 6 times SD. Today, 1080p HD is also popular which doubles the uncompressed bitrate required for 1080i. Alongside 1080p contribution, we are starting to see UHD services gain traction which carry a 12Gb/s uncompressed burden per service. Putting this into context, a typical requirement could be the need to contribute multiple UHD feeds from an event over 10Gb network infrastructure which is not feasible (even for a single service) without compression. This explains the gap that JPEG XS is designed to fill: The codec is not intended to focus on delivering particularly high compression ratios and in this regard, AVC, HEVC or even JPEG 2K will prove superior.
JPEG XS is designed to provide a useful compression gain (for example, 10:1 so the uncompressed bitrate is reduced by 1/10 which is sufficient to solve most contribution challenges within the premium content sector). Uniquely, JPEG XS can achieve this whilst preserving the essential qualities of uncompressed video (true lossless VQ and negligible latency) to a much greater extent than alternative codecs. With latency optimised HEVC or even JPEG 2K, latency will be in the order of 100ms or more; with JPEG XS it is likely to be sub 10ms. Additionally, JPEG XS offers seamless integration into new ST2110 workflows, as well as legacy compatibility with MPEG TS workflows when required.
Differentiating by codec implementation
The JPEG XS codec can be implemented either in software (running in CPU) or hardware (running in FPGA). Although the CPU option is important to support several use cases, such as JPEG XS decoding in public cloud, the FPGA approach is considerably more efficient. FPGA implementation achieves higher service densities at the lowest possible power consumption and is the approach used by Appear.
Although it requires a greater initial development effort, Appear uses FPGA and related ‘high efficiency’ technologies extensively to perform complex video processing within its portfolio. This ensures the end-to-end system benefits of using the most appropriate technology for the task are concatenated through the entire delivery chain, reaping significant energy and cooling savings that save customers money and benefit the environment. Additionally, efficient operation is also the key towards achieving the remarkable service densities Appear offers (per RU), and this philosophy is applied to all Appear codecs and solutions .
Appear implements its JPEG XS encoding and decoding solutions on software-defined, multi-purpose modules offering a choice of SDI or ST2110 interfaces for baseband video presentation. It is possible to maintain a single format throughout (eg. SDI to SDI or ST2110 to ST2110) or work cross-format (where encoder input could be SDI with decoder output presenting ST2110 into fully IP native studio infrastructure). Since JPEG XS functionality is provided as a modular option within the versatile X Platform ecosystem, you can dimension the platform optimally for your needs (up to an impressive 72 services encoded or decoded per 2RU) and integrate JPEG XS with multiple other functions, such as other codecs, within the same platform. The X Platform empowers you with the flexibility to choose exactly how the system is built, blended with other functions and dimensioned.
Differentiating by offering all encapsulation options
In addition to supporting SDI and ST2110 uncompressed baseband presentation in a flexible way, Appear supports all known standardised options for JPEG XS encapsulation.
JPEG XS to ST2110 encapsulation is commonly favoured within studio environments where the encoder will provide the compressed video layer encapsulated as a -20 essence, with pass-through audio and ANC metadata presented as -30 (or 31) and -40 essences. PTP is used for synchronisation and service discovery / network configuration can be implemented either via X Platform’s easy to use graphical user interface, JSON API or via NMOS*.
The ST2110 standard offers many distinct advantages within production environments but may not be optimal for all applications (especially Telco distribution). As an alternative, the VSF TR07 option encapsulates JPEG XS traffic into familiar MPEG TS, allowing all components (video, audio and ancillary data) to be packaged together into a single flow. This offers transmission using a format that is well understood and can simplify monitoring and redundancy handling. Using MPEG TS also removes any need to support PTP and can fit nicely with similar workflows based on JPEG 2K or even AVC / HEVC which are also encapsulated into MPEG TS. This option is often preferred for long-haul transmission over Telco networks.
Whatever encapsulation format you prefer, Appear has the breadth of portfolio to support it.
Differentiating by offering the lowest possible latency
Since all JPEG XS solutions are based upon a common codec, it might be assumed the latency performance of all JPEG XS solutions will be the same. This assumption would not be correct since beyond the codec, the encapsulation and stream processing elements are instrumental in defining the overall solution performance including end-to-end latency. For all encapsulation methods, having access to properly optimised, low-latency de-jittering and redundancy handling options is paramount.
For MPEG TS encapsulation, the list of bespoke customisations required to handle the JPEG XS format optimally deepens: Now it must be possible to reconcile PTS timestamps much faster than is required for other codecs, and all stages, including supplementary items such as FEC need to be optimised to work at much higher bitrates.
When designing its solutions, Appear concluded the only way to achieve optimum performance and realise the full low-latency potential of the JPEG XS standard was to create a new, purpose-built MPEG TS encapsulation stage using many years of experience building multiplexers to perfectly accommodate these differences. The new and fully optimised encapsulation stage enables Appear to deliver typical end-to-end latency figures in the order of 10ms which is remarkable and well beyond the capabilities of many other solutions.
Differentiating via superior integration
Appear’s JPEG XS solution benefits from being an integrated part of the X Platform ecosystem which offers the ability to support UDP, RTP, satellite, ASI, Zixi, SRT, MPEG TS, ST2022-6 and ST2110 workflows integrated within the same physical platform. Such flexibility means JPEG XS can become an integral part of virtually any solution and used with the comprehensive array of other codecs X Platform supports. As an example, it is possible to ingest a JPEG XS mezzanine feed and send it directly into an AVC / HEVC encoder module which can then provide a fully built ABR ladder as an output ready for packaging. Thanks to an aggressive development roadmap, even HLS and DASH packaging will feature as an integrated option for X Platform from Q1 2023!
Thanks to a fully modular design, X Platform can be scaled back to provide a compelling answer to simple, compact, low-cost solutions but will always be ready for massive expansion just by adding additional modules. Major broadcasters and telcos rely on X Platform because it is so scalable and flexible and have helped Appear make it one of the most operationally friendly solutions on the market today: X Platform can police traffic, manage advanced redundancy options, keep you informed with advanced monitoring and, thanks to the integrated firewall feature, keep your operations secure.
Whether contemplating a straightforward JPEG XS contribution / distribution solution or factoring JPEG XS as the heart of a wider, more complex ecosystem, the X Platform provides the ability to implement and scale the perfect solution for you.
Differentiating by offering added functionality
Appear’s JPEG XS portfolio is fully featured and has been designed to support essentials such as exportable video input and output confidence monitoring, ST2022-7 seamless network reconstruction, a frame resynchroniser option for decoding and Dolby E pass through using Dolby Safe mode. Both encoder and decoder also support auto input format detection for any supported resolution, which extends from SD through to UHD.
Additionally, services can be ‘time-aligned’ throughout the encode / decode chain (a critically important feature for ground to cloud and VAR applications) with mechanisms in place to support either ST2110 or TR07 encapsulation.
Appear JPEG XS solutions are natively transparent to many HDR formats and will automatically recover HDR metadata
Audio is currently passed through transparently, and in addition to offering the glitch-free assurance provided by Dolby Safe mode, Appear supports a much wider range of presentation flexibility than many competitors: For example, ST2110 audio can be presented from 1 stereo channel per essence to 8 stereo channels. During early 2023, Appear plans to add a variety of audio processing options, such as Dolby E & ED2 decode, to allow passed-through audio to be transcoded and presented in alternative forms directly from the decoder.
Appear is committed to provide the most feature-rich JPEG XS solutions on the market and intends to fully utilise the modular benefits of X Platform to continue adding to these benefits well into the future.
Conclusions
This paper hopefully shows that not all JPEG XS solutions are the same; there are many options, features and attributes that need to be considered and when contemplating TR07 encapsulation even fundamental properties, such as latency, can vary significantly between manufacturers.
Appear has the expertise and focus to excel within the JPEG XS solutions arena. If you have not considered Appear for your project, please put us to the test and challenge us to demonstrate how our focus and differentiation in this speciality can benefit you.
