T&M Merges Into QC

George Jarrett hears how the EBU, producers and smart vendors are perfecting automated QC, but contextual perception will always over rule technology.

Test and measurement, typified by waveform monitors and vector scopes, has been as ubiquitous to the VT era as Steenbeck editing tables were to film, but we are fast leaving the linear tape world behind and have adopted a file-based world with automated nonlinear workflows in QC. All the old technical measurements are now wrapped into a software process, and file analysis at faster than real-time will produce results that can be used to highlight particular things to watch for in eyeball reviews. The difficult trick is tuning everything to test the right things.
"T&M has really gone away. If it is not software it cannot handle files," said Andy Quested, chair of the EBU QC group. "Nobody is building in T&M test rooms, except possibly around the archive. Why use scopes? A waveform monitor has a human interface. It measures and reads, but it is dead easy for a machine to say 1.2v is too big."
The EBU QC project has certain strategic programs. "To do that we need to know what QC means exactly," said Quested. "The first job was to define QC across program genres and functions."
Step one was bringing together broadcasters, content producers and QC device vendors in a bid to identify every test you could possibly do to a piece of content. In the UK, The Digital Production Partnership (DPP) has already created a sub set composed of 40 of the EBU tests.
With a file you can’t physically see anything and the advantage is that you just need something to examine it and say it will play back in compliant devices. Because of interest in insert editing, the EBU split the QC development process into two primary elements.
"Is the program OK? Captions OK? It plays OK? You have put everything in? That is the first," said Quested. "With QC devices, it looks to see what the waveform monitor would have looked at before. If you have made a mistake, it is here that you can replace anything. QC is about too loud, too bright, too big, and basically it is the same as the technical reviews for VT. At the point of file delivery QC will be, is the format correct? Is the metadata correct? The tracks correct, the timecode correct?
"The biggest issue is if you want to change something after the file is delivered and set for TX. With a tape, yes, you can go back and drop an element in or change something," he added. "Do you do that or produce a new generation? If it was brought back for a swear word, you may want the first version held for after the watershed. The same applies to a file – but not always a drop in. It is going to be a re-make. Put this into context: if you need to drop things into files a lot of the time you have a poor production process of poor workflow. The emergency edit, the last resort, is not to be encouraged as part of general workflow."
The EBU has numbered every test, and there are 160 of them.
"It took six months to get every test listed, and we had 500 submitted. We looked at these and eliminated duplicates," said Quested. "We had problems with misplaced naming conventions as well, but the group pulled the 500 down to under 200 by aligning the names. All aliases sit behind each name for every test.
"It is about levels," he added. "You do not have to go from traditional to automated in one step. People ask for scopes on the front of QC devices, but why? Traditional waveform does not tell you enough. The only reason to use it is that it does more than one job."
The R 128 loudness specification has been adopted across most of Europe, and the QC tests are likely to follow. But above the technical and objective elements of QC there sits the subjective.
"The arguments and debates I get involved in (at the BBC, where he is head of HD and UHD technology) are usually about artistic interpretation, or the use of material there is no way of repairing," said Quested. "Technology can always be over ruled by editorial. It should always be that way because the best stories in the world usually come from mobile phones. It is about context."
The EBU QC corrections roster is effectively a reference source, and the DPP sub set of tests along with the producer checklist for file delivery it has created typify what broadcasters and producers can extract. What then do vendors have to say?

The Digimetrics take on QC
Eric Carson is general manager, founder and chief product architect for Digimetrics, and he attends the EBU QC meetings. He serves the group as an editor for several of the recommended items.
Asked if T&M is merging into QC, he said: "Linear T&M still serves a role in broadcast play out centres and MCR in several regions around the world.
"However, the reliance on linear T&M is diminishing. In the case where regions have migrated to file-based interchange, the need for linear T&M on those file-based deliveries is essentially moot, since the file metadata can, with additional shims, like AMWA and DPP have defined, properly describe the container and essence, and the baseband equivalent essence components will exhibit different artefacts, when present, than what linear T&M is designed to catch," he added.
"For example, analogue overshoot/ undershoot doesn’t exist in a file-based delivery, because the quantiser clips analogue values during digitisation," he continued.
"Additionally, transcoders can introduce artefacts that, while totally legal from a linear T&M perspective, significantly decrease the user experience (such as macro blocking noise, digital hits or incorrect audio track allocation)."
Has the EBU identified all the tests that can feasibly be done, and what role is SMPTE playing?
"SMPTE really hasn’t been involved in this effort, yet. Ideally, the EBU would submit this work to SMPTE for consideration as a recommended practice or at least a registered disclosure document when we are complete in the definitions," said Carson. "As far as the definitions go, I think we have defined over 90 percent of the tests that can feasibly contribute positively to an automated workflow, whether that workflow is broadcaster interchange, archive digitisation, or pre-flight distribution/play out. There are some unique or vendor-proprietary items that we specifically have avoided defining because they do not appear to either help an automated workflow (i.e., they require just as much human intervention/review as a linear T&M process) or because they have dubious capabilities at the present time," he added.
Does membership of the QC group help in terms of identifying development opportunities?
"This really hasn’t come into play yet. I think all the vendors that participate have had a pretty good cross section of the tests we have defined. Certainly, some vendors are stronger in certain testing areas than others, but this is simply good market dynamics at work. There does not appear to be a single vendor today that covers all of the testing defined by the EBU. Today’s automated QC vendors are competing on features other than test definitions and methodology; they are competing on performance, accuracy and integration," he added.

here.
(IT)