Editor’s Note: This content is contributed by Antonin Descampe, Compression technologist at intoPIX
What is JPEG XS and how does it differ from JPEG 2000, Motion JPEG, or other MPEG standards? What about its implementation in FPGAs?
The main difference between the new JPEG XS standard and existing codecs as developed by JPEG, MPEG, or other standardization committees is that compression efficiency is not its first target. While other codecs primarily focus on high compression efficiency, disregarding latency or complexity, JPEG XS addresses the following question: “How can we ultimately and reliably replace uncompressed video?” JPEG XS handles increasing resolutions, frame rates, and a number of streams while safeguarding all advantages of an uncompressed stream. JPEG XS also delivers:
Low power consumption in FPGAs
Low latency in coding and decoding
Ease of integration
Extremely low resource usage on FPGA devices
And there is no need for external DDR memory! These are major advantages compared to other compression technologies that target FPGAs.
IntoPIX JPEG XS
No other codec fulfills this set of constraining requirements simultaneously. Therefore, JPEG XS is a strong challenger of uncompressed video in every aspect, with the additional advantage of significantly reducing bandwidth and video data. Users who are handling high data rates in their video systems can now use the intoPIX JPEG XS mezzanine codec as a cost-efficient solution to ease up the video flow through their designs.
What makes JPEG XS the go-to mezzanine compression solution for Xilinx users?
We have already mentioned some of the main advantages of a JPEG XS implementation in Xilinx FPGAs, e.g., a much lower complexity than any advanced MPEG inter-frame codec, hence cheaper FPGA implementations, and a tiny footprint with no need for external frame storage in DDR. This also leads to a huge difference in terms of power consumption. JPEG XS consumes only a few lines of pixels to compress video. It has a microsecond-latency and can thus be run throughout a whole live workflow without inducing the latency of even a single MPEG-4 AVC / H.264 encoding-decoding step.
This achievement did not happen by accident. The optimized FPGA implementation has been a focus from the early stages of the development of the JPEG XS algorithm. The team included FPGA developers, who were actively involved in the engineering of the new algorithm.
Xilinx FPGAs and SoCs are used in the intoPIX cost-effective solution—leveraged in various video workflows such as broadcast, autonomous driving systems, KVM extension, VR/AR gear, and wireless. A JPEG XS implementation in those Xilinx designs further improves the cost reduction and offers new development possibilities: more streams, higher resolutions. The JPEG XS has been ported by intoPIX on almost every available Xilinx FPGAs and SoC, including Spartan®-6, Spartan-7, Artix®-7, Kintex®-7, UltraScale™, UltraScale+™, and Zynq® UltraScale+ devices.
What sort of compression is achievable with JPEG XS and what are the compression choices for a 4K video with JPEG XS?
In a nutshell, the typical operating point for visually lossless quality with JPEG XS is a 10:1 compression ratio.
However, it is important to take resolution and content type into account when identifying a maximum compression ratio. For instance, natural contents, e.g., video camera outputs, usually reach higher compression ratios than artificial contents, e.g., PC desktops, for a given quality level, due to the higher correlation between neighboring pixels.
Moreover, “visually lossless quality” can be interpreted differently based on end-user quality standards. During its development, JPEG XS was tested against the strictest quality assessment procedures, ISO/IEC 29170-2: “Evaluation procedure for visually lossless coding,” the threshold that guarantees an “indistinguishable flickering” between original and compressed images often referred to as “visual transparency.”
JPEG XS tests were performed on an extensive range of contents: screen content, Computer Generated Images (CGI) and natural imagery, and mixed images. The results are available in the following table. The lower range of compressed bitrate in the table typically targets natural content, while the upper range is aimed at more complex content or use-cases requiring full visual transparency.
JPEG XS Test Results
We hope this will have given you a better understanding of the JPEG XS technology and its advantages for Xilinx FPGA users.
IntoPIX is an IP provider and Certified Alliance Member of Xilinx. Please feel free to contact intoPIX if you want more information about the JPEG XS technology. We would be happy to talk with you.