Codecs & workflow: Making the optimum choice. Part 2 - delivery

In part one, I described how tapeless cameras offer a wide choice of codecs, with some also having the ability to output raw and/or uncompressed files. The next step in the workflow is to choose an edit codec and then a delivery codec. The choice of edit codec should be influenced by the delivery codec in order to maximize the video quality.

In the days of videotape, Digital Betacam was a popular delivery format for standard definition (SD) content. For high-definition programming, D5-HD was a popular choice. Now HDCAM SR is the favored format specified by broadcasters for program delivery. With 12 audio tracks, it provides flexible support for surround sound delivery.

Video Codecs

As broadcasters move to file-based operations, they are beginning to specify the file formats they want used for program delivery. For SD delivery, the D10 format is popular, using MPEG-2, I-frame 4:2:2 coding at 50Mb/s (the wrapper can carry up to 8 AES audio channels.)

For HD, there are more choices. Again, 50Mb/s is a popular data rate, but for HD long GOP coding must be used to handle the increased resolution. The essence codec specified is that used by Sony’s XDCAM HD: MPEG-2 long GOP at 50Mb/s or at lower rates of 25Mb/s or 35 Mb/s. Other broadcasters will accept edit codecs like DNxHD and ProRes.

The delivery formats get more complex when audio, closed captions and metadata are included. One broadcaster may want a stereo full mix, another may want 5.1 music and effects (M&E) plus a 5.1 full mix as well as stereo M&E plus full mix. Each broadcaster may have their own house track allocations. Multi-format distribution adds further options.

When videotape was delivered, it usually included a record report or similar, a mass of metadata about the contents. The cassette and box also carried labels, with basic metadata, tape number, program title, audio track layout, etc. Files may or may not include metadata, but so far there has been little standardization.

This all adds up to an element of chaos. A post house must render the final cut in all manner of formats depending on the customer. This means saving a number of preset templates in the editor to suit each client and the destination. Metadata will also need to be different for each client.

There is a way

In the UK, the main broadcast networks took the route of cooperation to create a single delivery specification that would cut through all the confusion. Under the auspices of a trade body, the Digital Production Partnership (DPP), they worked closely with the Advanced Media Workflow Association (AMWA) to create an application specification for a file delivery to broadcasters and playout facilities. The specification, MXF Program Contribution — AS-11, will define program file format to replace delivery on HDCAM SR tapes, and will be in full use by 2014.

In view of the timescales, and to be future-proof, the DPP decided to specify an MPEG-4 codec rather than the MPEG-2 in common use today. The chosen format, AVC-Intra Class 100, High Intra 4:2:2 profile at level 4.1 (as defined by SMPTE RP 2027:2011), retains a higher picture quality further along the workflow to transmission. The preparation of programs for transmission may need the creation of promotions, and the 10-bit coding of AVC-I gives cleaner compositing for any graphics that may be added.

It is often said that AVC needs a more powerful workstation for editing than MPEG-2, bu,t by 2014, Moore’s Law should make such restrictions on codecs unnecessary. Plus broadcasters may well have to upconvert HD to 4K in the future. Who knows?

Delivering as AVC-I does not mean capture and editing need be as AVC-I. A typical workflow could be shot ARRI raw, edit as DNxHD and deliver as AVC-I. The choices lies with the producers; the only restriction is that for DPP members, delivery should be AVC-I.

Outside the UK, there does not seem to be the same call for AVC-I, and long GOP MPEG-2 is very popular. AMWA’s AS-11 includes all the codecs of the AS-03 specification, (MPEG-2 MP or 422P, or H.264, any GOP structure, at bit rates of 5Mb/s up to 50 Mb/s) as well as D10 for SD files.

The complexity for delivery is not so much the codecs (the issue with acquisition), but in the wrapping – the audio, and the metadata – but that is for another article.

As tape is used less and less to deliver programs, and files become the norm, the custom and practice will develop. But, the use of standard specifications based on international standards like ISO/IEC MPEG and SMPTE provide a route through potential chaos.