Following my first post on the topic of HDMI distribution over IP and my subsequent visit to InfoComm 2016, I can confirm that HDMI over IP is most definitely here. There is a lot to choose from in fact! Here are some things things you should be aware of.
Many of the established names in AV distribution were displaying their new IP based HDMI distribution systems at the show. Given the fact that the enabling technology is now on the Tx and Rx side of a commodity network switch, and an expensive custom developed signal backplane is no longer required, the barrier to entry for this type of system has been significantly lowered for manufacturers, and therefore quite a few vendors that did not offer HDMI distribution products before have now jumped on this bandwagon. This is an interesting development, as it increases the competition in this space… almost always a good thing for the end user.
In my previous article, I mentioned two ‘form factors’ of these systems that I expect to see evolve. The first type is what everybody is offering now. Tx and Rx boxes that connect to a third party network switch. Now as the dust settles and the technology becomes more commonplace, I expect that vendors will start to try and distinguish themselves from their competitors. This is when I expect we’ll start to see the second form factor of product start to evolve (higher input density devices on the Tx side). I hope this happens quickly to be honest, because for systems with a large number of sources in the same physical space, installing all those separate Tx boxes is a bit cumbersome.
Now once again I need to come back to the distinction between uncompressed and latency free HDMI over IP, and well, everything else. This is an important distinction when deciding what product to implement.
When it comes to network bandwidth, there are two classes of distribution systems available at the moment: those that allow you to move the video over a 1Gbit network, and those that require a 10GBit network. Practically all of them being marketed “visually lossless” and “low latency.” Given the fact that even the lowest grade of 4k video has a raw bandwidth of well over 1Gbit/sec, to get these signals onto the network, the 1GBit solutions are having to apply heavy compression using video codecs such as H.264, H.265 (HEVC), VC2 and JPEG2000 (J2K). Now I won’t go into detail about these codecs, but I will stress that they are all what is referred to as “lossy,” i.e. not lossless. There is always a sacrifice of quality with these compression formats. The term “visually lossless” is of course a marketing term, and whether or not the compression is noticeable to a viewer depends on visual acuity of the viewer, knowing what to look for, the type of content, the viewing distance and environment, screen size and quality, etc. All of this is not to say that these systems are bad or shouldn’t be used. There are absolutely scenarios where they would be perfectly adequate, so if they meet your needs, then by all means use them. However, If you need to move high resolution graphics and images around, please be aware that ‘visually lossless’ quickly turns to ‘clearly lossy,’ therefore I wouldn’t recommend these systems for that type of application.
In terms of latency, it varies quite substantially between the Gbit solutions, anywhere from 20-300ms. This is also influenced by features like scaling. If latency is not a concern for you, then that’s fine, but where real-time is a requirement, or where there is user interaction, like in a KVM or remote control scenario, this can be problematic.
For the solutions requiring a 10Gbit network, there are two sub-classes. The first applies compression to all sources using a lighter implementation of JPEG2000 or a newer format called TICO. Very low latencies can be achieved with good implementations of these technologies, with TICO even managing sub 10 microsecond latency. The second sub-class does not apply compression unless it is absolutely necessary to keep the video within the bandwidth constraint.
It goes without saying that less compression results in a higher image quality, therefore these 10GBit solutions generally produce a better quality picture than their 1Gb/s counterparts. Still, these compression technologies are more suited for video content, so for graphics and KVM type applications, the compression artefacts can be (clearly) visible.
The second subcategory (no compression unless called for) is sending truly uncompressed HDMI over the network unless the bandwidth of the incoming signal exceeds the network capacity. In this case, it applies a proprietary compression format that is mathematically lossless. What appears on the display at the other end is therefore an exact representation of the incoming signal. When it comes to latency, this product measures in single digit microseconds even with scaling and compression active, which is completely imperceptible. This makes it suitable for even the most critical of applications. This is really the only class of product that fits into the uncompressed and latency free category that I wrote about before.
There are multiple variables that you may want to consider when evaluating distribution systems, but there is one more that I’d like to mention specifically: HDCP handling. Because went it comes to this, not all systems are created equal.
First it’s important to understand that in order to process the HDMI signal and manipulate it (scale, transcode, compress etc), HDCP needs to be removed as the signal comes into the system. This is standard practice. On the receiving end, HDCP can be reapplied before the signal is put onto the HDMI cable, ensuring that only validated sinks can play back a source. Not doing so would very likely result in a court case somewhere down the road, so I expect most manufacturers to this. More interesting is what happens as the signal traverses the network. HDCP makes use of (AES128) encryption to make sure that the signal cannot be intercepted and copied anywhere along the signal path. Not all systems out there encrypt the IP-encoded signal as they push it onto the network though, meaning that the signal may not be protected along that part of the signal path. How the HDCP body feels about this I don’t know, but it may well be a concern for companies pushing copyrighted video or video containing sensitive information across the network. I’m told that there are manufacturers that make the end user accept liability for HDCP non-compliance in their product, and I’m sure there will be others who are not compliant and don’t even inform their users. Just something to be aware of.
To sum up, there are now many offerings available to anyone interested in distributing HDMI on a standard IT network. These products can be split into a few categories in terms of video quality and performance, and they all have their own pros and cons. It is up to you to evaluate these systems according to your own requirements, but with this article I hope to have at least armed you with some additional knowledge to help you ask the right questions and narrow things down for your own applications.
Originally published on by Edwin Edelenbos on LinkedIn.