What is Motion JPEG 2000?

You’ve seen the technology in use with many AV over IP video systems, but what exactly is Motion JPEG 2000?

We’re familiar with JPEG images, found all over on the Web and on our phones and tablets. JPEG was created in 1992 by the Joint Photographic Experts Group, who followed up with JPEG 2000 in Y2K. As a new standard, it was glorious. Also known as JP2, advanced wavelet encoding could save images in a lossless or lossy version. While JP2 offered improved performance, it never gained acceptance – the difference wasn’t enough to challenge the popularity of JPEG.



ve seen the technology in use for many AV over IP video systems, but what exactly is Motion JPEG 2000?

You’ve seen JPEG images, found all over on the Web and on your cell phone. JPEG was created in 1992 by the Joint Photographic Experts Group, who followed up with JPEG 2000 in Y2K. As a new standard, it was glorious. Also known as JP2, advanced wavelet encoding can save the images in a lossless or lossy version. While JP2 offered improved performance, it never gained acceptance – the difference wasn’t enough to challenge the popularity of JPEG.

Going Into Motion

Acceptance was much improved when the group upgraded the legacy Motion JPEG to use JP2 technology. A Motion JPEG 2000 video stream is composed of a series of JP2 images at 24, 30, or 60 frames per second. The first adopters were archivists like the Smithsonian, who discovered they could compress videos at a 3:1 ratio without losing original quality. Video editors preferred MJP2 over MPEG, because it was easy to cut between individual frames. When you go to a digital movie theater, you’re watching a high-end version of MJP2.

Adoption for network streaming was a different story, as MJP2 streams are massive, and MPEG H.264 performs better for highly compressed streams. However, with the advent of dedicated AV 1G/10G networks, MJP2 has found a new life for AV over IP, as:

  • Video can switch between frames, assuring zero latency
  • Compression shrinks bandwidth to fit the network – 3:1 compression for 10G, 20:1 compression for 1G networks
  • The codec is free, with zero royalties

Motion JPEG 2000 by Any Other Name

Some products will list MJP2 by name, others may use the codec under a brand name, and some use a variant of MJP2. I suspect branded codecs are really Motion JPEG 2000 – what vendor would invest a million or so for a new codec that works the same as the free codec?

  • Crestron NVX Pixel Perfect – MJP2 with tweaked settings
  • Extron PURE3 – it’s not a patent, but a trademark for sending compressed streams (likely MJP2), audio and control through a network
  • SVDoE – “Proprietary” but compression ratios like MJP2
  • Atlona VC-2 – Britain’s BBC’s Dirac codec is similar to MPEG but uses JP2 instead of JPEG for the I-frame in the GOP. Dirac Pro (SMPTE VP-2), designed for broadcast production, can handle 4-8K video and encode video with only I-frames, essentially the same as MJP2.

Video in Transition- Hauling a 4K load Over a 1080p Bridge

New technology brings new challenges and solutions in AV design

The transition from 3:4 analog to 16:9 digital HD was reasonably painless. For meeting rooms and living rooms, HD screens had had little impact on room design – they were about as tall as 3:4 screens, with more horizontal real estate.

Converting analog video to HD was a greater challenge for twisted-pair video distribution. “Video Voodoo” problems arose as the precise kind of wire twist affected distance, resolution and performance. Those problems faded away with the introduction of HDBaseT, a chipset that converted digital video into packets. Think of HDBaseT UTP cable as an 8G bridge, easily transporting 3G 1080p video, audio, control data and Ethernet for up to 300 feet.

Enter 4K, an easy step for TV vendors, not so easy for twisted-pair video. That 8G bridge now has to carry a 10-16G load, more with 4:4:4 color and 10-bit HDR. Not happening, so something has to change – lighten the load, strengthen the bridge, or a bit of both.

Building the Bridge with PAM

PAM stands for Pulse-Amplitude Modulation, that expresses data using amplitude in a series of signals. PAM-5, used by 1G Ethernet uses values of -2, -1,0,+1 and +2 VDC to represent bits of data. Each of the four twisted pairs carry up to 125 mbps, summing up to 1G of data. 10G Ethernet, both HDBaseT and IP Switching, employs PAM-16, encoding 16 levels with added physics. The limiting factor in PAM technology is RF noise, as it prevents the receiver from sensing the right levels. This is a key factor in 10G Ethernet, as the level steps are much smaller. For this reason, shielded cable should be used for all 10G applications. 2.5G/5G Ethernet uses a lighter version of PAM-16 designed for Cat 5e/6 UTP applications. Of course, optical fiber is always the best carrier, as there is no RF noise to interfere with the signal.

It’s interesting to note that AV platforms we view as different – AV over IP, QAM digital cable, 8-VSB off-air digital channels, and HDBaseT – are all based on variations of PAM technology.

Compression – Lightening the Load

Full 18G 4K60 HDR video can’t travel over a 10G bridge, so the stream has to be compressed to fit. There are two technologies in use today:

  • VESA Display Stream Compression (DSC) in newer HDBaseT systems, creates a visually lossless stream, typically at a 3:1 ratio, reducing an 18G 4K60 HDR stream to 6G with little loss in quality. Examples include Crestron DM 4KZ and WyreStorm 18G HDBaseT systems.
  • Motion JPEG 2000 was designed for video storage, reducing files by 3:1 without losing original quality. With today’s higher bandwidth, the codec has a new application in AV over IP technology:
    • 10G IP Switching. Employs 3:1 compression to deliver mathematically lossless video. In use today in SVDoE and Extron systems.
    • 1G IP Switching. Employs 20:1 compression to deliver visually lossless streams over 1G Ethernet networks. Available from Crestron, WyreStorm, Extron, Kramer and many others.

VESA DSC has a key advantage – compression only affects streams starting at 4K60 4:4:4, other streams are uncompressed.

The good news is there are solutions in place for the challenges in 4K video distribution.

  • Sites with existing Cat 5e/6 UTP wiring can transport 4K video using 1G IP switching gear, and companies like WyreStorm and others have HDBaseT extenders that can transport 4K video over Cat 6, with some resolution and distance limitations.
  • New installations have many options for 4K distribution over Cat 6a/7a STP cable or fiber.

AV Over IP – A Primer

AVoverIP

AV Over IP is the term for technology that delivers audio visual content over Ethernet. The term also implies that content traditionally sent or switched over analog or digital switching now employs IP packets and standard Ethernet switches between the source and destination.

There are two basic AVoIP applications, Distribution and Presentation.

Distributed IP Delivery

  • Content is distributed over a large area
  • Endpoints described as encoders and decoders
  • Usually a high ratio of decoders to encoders
  • Streams are highly compressed to save network bandwidth
  • Half- to 1-second latency is acceptable, depending on application
  • Streams can provide captioning data and audio sync

Distributed content is sent over large areas, including nationwide through Netflix or over a single site from cable channels, modulated over RF coax, or via an IP network. Due the larger scale of distribution, streams should be as small as possible, able to carry ADA captioning, and don’t require instant switching for viewing. The standard format for this application is MPEG, usually MPEG2 for off-air TV and in-house RF channels, and H.264 for commercial and consumer IP-based TV.

MPEG is designed for maximum compression – H.264 can easily compress a 3Gbps (bits per second) 1080p video to a 15 mbps stream, a 200:1 ratio, and consumer streams are compressed far more. The secret sauce is called the GOP, the Group of Pictures, and only the first frame is an actual image. MPEG compresses the first frame into an image similar to a JPEG. The encoder then captures a few more frames and notes objects that move or change color, and saves just that data. This is called inter-frame encoding, as the encoder is continually cross-referencing frames. When a decoder changes a stream, it has to capture the first several frames in the next GOP and rebuild the video back to its original content. This deconstruction/reconstruction process takes time, creating about a half- to one-second delay. That’s why you experience a pause when changing MPEG channels.

Presentation IP Delivery

  • Displayed in a defined area
  • Endpoints described as transmitters and receivers 
  • Simpler to define and expand input/output configurations
  • Large streams require a dedicated IP network
  • Instant switching, about 2-4 frames
  • Streams composed of images and audio tracks, no captioning data

When a system is delivering a live presentation or event, the content on the video screens must to be in sync with what’s on the stage or podium, and cameras need to be changed instantly. Visible latency is disconcerting to the presenter and audience. This is the traditional application for AV switching systems.

The new counterpart to AV Switching is routing content over an IP network. Instead of a central switch with multi-port input and output cards, designers can define any number of individual transmitters and receivers, routed through a standard Ethernet switch.  Video frames are converted to stream packets using Motion JPEG 2000, typically compressed to 3:1 to 20:1 ratios. As the stream consists of individual images, switching is fast, with little latency. 

However, latency is a bit looser than AV’s vertical-interval switching. IP guarantees delivery, but there is no central clock to lock down timing, so there are network variables that could affect latency. As with AV switching extensive scaling at the input and output points can affect latency as well. As there is not sync data as there is in MPEG, audio could be behind – or ahead of the video, especially if the video is highly compressed and scaled and the audio just passed through. I would imagine some IP systems offer settings to minimize audio timing issues. 

What is the Effect of 4K video on commerical IP distribution systems?

For commercial systems, 4K is more of a marketing pitch than a reality. Current room and screen designs can’t take advantage of 4K, and very little content will be 4K for some time. It just isn’t needed. 

MPEG distribution technology is typically limited to 1080p. However, H.264 does support 4K and “zero latency“. For the present, 4K codecs such as HEVC (H.265) and VP9/AV1 aren’t usable for commercial applications.

 

Bridging the 10G/1G Divide in IP Video Switching

1G-10G Bridge

IP Switching Tech Simplified

Up to now, AV over IP solutions are divided into two groups – 10G or 1G networks. While both market their solutions differently, they use the same technology. Under the hood, all use Motion JPEG 2000 or a variation, developed years ago, initially for archiving video. Files can be compressed up to 3:1 without losing original quality, and the video stream is composed of individual compressed frames, simplifying video editing and switching. When you go to a digital movie theater (and almost all are now), you’re watching a 4K Motion JPEG 2000 video.

AV vendors quickly moved to JPEG 2000 for IP switching for the same reasons – excellent compression, switching, and 4-8K capacity-and the codec is free. While many vendors have tweaked and renamed the codec to make it proprietary, the functionality is the same.

  • 10G systems use “mathematically lossless” 3:1 compression to reduce a 16G 4K video to less than 6G. That’s great, but 10G switching can’t run well over Category cable, requiring fiber for all connections.
  • 1G systems use 20:1 “visually lossless” compression, resulting in an 800 Mbps stream that can use standard 1G switches and Cat 5e/6 cable. It’s a good tradeoff – 4K quality is still great and more functional, as most commercial systems will be routing 1080p streams, anyhow.

The key point is that the divide between 10G and 1G systems is purely arbitrary – the only difference is compression.

2.5G/5G Networks over Cat 5e/6 Cable

The division between 10G and 1G systems will be narrowed with the advent of 2.5G/5G Ethernet. While 10G is useful for connecting switches, it’s bust for wiring PCs. Sites already have millions of miles of 5e/6 Cat cable installed; there’s no way they will replace wiring with Cat 6a/7 or fiber.

2.5G/5G technology isn’t a new or proposed standard, it’s always been a part of 10G technology. In basic terms, 10G is a large checkerboard of data, while 2.5/5G is a smaller section in the center. That smaller section enables the transmission of 2 to 5 times more information over Cat 5e/6 cable than 1G Ethernet.

What’s new is the advent of multi-gigabit 1/2.5/5/10G IP switches and IP ports than can portion out the right speed for the application – typically 10G between switches, 2.5G to Cat 5e feeds, and 5G through Cat 6 wiring. It’s a great solution, adding new power to existing networks just by upgrading switches instead of installing new copper or fiber.

While a growing number of multi-speed switches are available, PCs, laptops, IoT devices and AV over IP gear are supplied with 1G ports. That will change over time as IT admins gradually adopt the new capability.

However, change may be quicker in AV over IP solutions, as suppliers can go beyond the 1G barrier, delivering 4K video at two to four times less compression over standard 5e/6 Category cable – just by upgrading their Ethernet ports. That’s a game-changer!