Four weeks ago, I posted in this blog about the IRT MXF plugfest, the new MXF profiles that were published in Germany this year by the ARD and ZDF, and how these new profiles would bring forth a new era in interoperability. This week, the first results of that plugfest and reaction from some of the end users and vendors were presented at a conference on file-based production also hosted by the IRT in Munich.
 
As usual, the results were fascinating. As with all statistics, they could be manipulated to back up any point you wanted to make, but for me there were a couple of highlights.
 
First, as mentioned in my last post, this was the 9th such MXF plugfest, and therefore we have a good historic dataset. Comparing previous years, there is an obvious and steady increase in both the interoperability of exchanged files and also compliance with MXF specifications. For most of the codecs and variants tested by the 15 vendors who took part, over 90% of files are now exchanging successfully (up from 70-80% five or more years ago). In one case, the new ARD-ZDF MXF profile HDF03a, 100% of the files submitted interchanged successfully. Quite interestingly, the same files all failed a standards compliance test using the IRT MXF analyser.
 
This highlights one of the difficulties the industry faces today with file interoperability, even with constrained specifications such as the AMWA Application Specifications and ARD-ZDF MXF profiles. The IRT MXF analyser categorises test results as pass, fail, or with warning. It is notable that all files with MPEG 2 essence (e.g. XDCAM HD) either failed or had warnings, while AVC-Intra and DNx files each had a significant number that “passed.” However, when it came to interoperability, the differences between the different codecs were much less obvious. One theory would be that because MPEG 2 in MXF is the oldest and most widely used MXF variant, it has resulted in a near de facto standard that enables a reasonably high degree of interoperability – despite the fact that most of these files are not compliant with specifications.
 
I mentioned in my previous post that the new ARD-ZDF profiles have accommodated this deviation from specification in legacy files by specifying broader decode parameters than encode parameters. This was the focus of my presentation at the conference this week, illustrated through the use of children’s toys and the game of Pictionary. However, the additional decoder requirements specified are not without issue. For example, if not impossible, it’s certainly impractical to test all the potential variations covered by the broader decoder specification given that it would be difficult to find test sources that exercise all the possible combinations of deviation from the encoder specification. In another area, while the profile says that the decoder should be able to accommodate files with ancillary data tracks, there is no guidance as to what should be done with the ancillary data, should it be present. As a vendor, that’s particularly problematic when trying to build a standard product for all markets where the requirements in such areas may vary by region.
 
Overall though, while there are improvements that can, and will, be made, it’s clear that for vendors and end users alike the new profiles are a big step forward, and media facilities in Germany are likely to rapidly start seeing the benefit in the next 6-12 months. Exciting times lie ahead.