TVB Europe White Paper

							                        TVB Europe White Paper
                    HD Playout – Time to Outsource?


In February 2006 Red Bee Media announced that it will playout the BBC’s HD trial
this summer and is also offering broadcasters the opportunity to trial their channels in
HD. In this white paper we will address how Red Bee Media researched and built this
HD playout solution and how we have developed various levels of end-to-end HD
solutions for a variety of broadcasters.

Red Bee Media has spent the last two years testing HD workflow, media
management, playout architecture and distribution systems. Starting during IBC2004
using, what was then, cutting edge Mpeg2 HD encoding for both server and satellite,
then in 2005 for four months simulcasting using Mpeg 2 and the newer Mpeg4-AVC
(also known as H264 or Mpeg4-part 10) satellite encoding. This also included testing
DVB-S2 satellite modulation. These tests were to compare distribution bitrates, set
top boxes and the viewer experience.

Although both of these test channels were unencrypted and ‘the first high definition
channels to be broadcast from the UK’, we specifically put them on satellite positions
that made them unavailable for a mass audience. This meant we had access to a
much wider set of real content, with no distribution rights issues, allowing us the
freedom to properly test the boundaries of the technologies. This did present a
downside at IBC 2005, where some visitors to our stand did not realise we were
showing a true end-to-end channel transmitted from London, rather than locally
generated content. When viewed on a 45” domestic display, this was obviously not
quite as good as the uncompressed servers/graphics devices on broadcast quality
monitors as seen on neighbouring stands – 1.5-3 Gbs compared to 14-20Mbs !

During this time we also did a number of demonstrations for small to large
broadcasters looking to launch HD and hosted the first Omneon / Miranda /
LinearAccoustic HD Technology sessions – an extremely detailed and useful event
for technical and operational management as well as channel owners.


Acquisition, Production and Delivery

One of the most important aspects of launching an HD channel is understanding the
workflow – from the very start of content life until its last repeat on whatever platforms
it is transmitted. Both channel owners and broadcasters must understand where
quality can be affected, and try to prevent this happening.

HD can have a very large acquisition datarate from devices like cameras and
graphics equipment, but this is usually much lower when recorded onto tape or disk
(depending on format). That gives the first compression level, which is ‘lossy’ and
therefore restricts the amount of generation drops that can be performed. (SD has a
much lower ratio between originating datarate and that recorded onto tape, meaning
more generations are possible before it becomes apparent.) The result is the
beginning of ‘concatenation effects’ – something that can start well before purchased
content even reaches the channel owner or broadcaster. The effects of this may not
actually be seen until much later in the process, but by then it may be too late or too
expensive to correct.
The frame rate and line rate conversion processes are the next problem areas,
where material purchased from overseas may need to be converted to a new
distribution rate. This is where editing, reformatting and compliance will be performed
and promotions can be made. At this point it is important to consider the quality of the
converter and whether there will be another compression stage.

Up-converting of archive SD content may also be required. Again, consideration
needs to be made as to what grade of converters should be used, but more
importantly, what the content is and how it was originally produced. Red Bee Media
has used a variety of upconverters ranging in cost from £2k to £50k. Our experience
has shown us that upconversion should be monitored very carefully. Decisions on
what SD content to upconvert and transmit on an HD channel, as always, needs to
be made based on the topicality and relevance of the content, but also with the
quality of the result.

At this point in the process, subtitle and audio files may be converted, bearing in
mind the differing timecodes for different formats. Here, there is another potential
pitfall – some scheduling, playout, media management and server systems work at
up to 30 frame resolution, whereas you may be broadcasting at 50 or 60 frames per
second. It needs to be considered if software will be upgraded by manufacturers, if
playout providers will create workarounds or if frame-accuracy will be less important
to the channel owner.

When delivering content to the playout and media management provider, the tape or
file format used needs to be carefully considered, as there are many different options
presenting a balance between cost, volume and quality. Can your provider cope with
all of the above format conversion and editing processes to ensure the quality stays
as high as possible throughout, or does content have to be converted by the content
source?

File delivery is not technically difficult but the logistics are likely to be a challenge for
the sender. Tape formats and volumes will be the driver for file delivery, with
connectivity and network security also being expensive. Volume and the economies
of scale are the key to keeping the costs down.

With delivery tape formats there is again a cost and volume equation, but also the
same technical considerations as previously mentioned – what compression levels
are acceptable, what do your providers already use and - for channels with less
content - can you ratecard usage of facilities rather than have dedicated usage.

There a varying views on what formats to use. We are not going to get into the
720p/1080i or the ‘film look’ against the ‘natural look’ debates – very experienced
people have clear but opposing views. It is a subjective argument and very difficult to
see a right or wrong to either side. This goes back to the workflow planning stage
mentioned above - and until/if ever we can get to the goal of actually transmitting
1080p, and probably only then by satellite and higher bandwidth distribution
platforms, the channel will be a single standard only (unless encoder and set top box
manufacturers develop clean live-switching within a channel). The format decision
will probably be dependant on the content and acquisition process, rather than the
slight saving on bandwidth that 720p may offer.

Media Management and Playout

When thinking about playout and media management consideration needs to be
given to the resilience levels the channel owner wants. This is a balance between
cost and viewership, but it must be kept in mind that equipment is still new and
somewhat unproven compared to where we are with SD. Our tests with various
manufacturers over the last couple of years have helped us identify missing elements
and understand some of the pitfalls. In turn, we have been able to provide feedback
which will ultimately provide us with more functionality and features.

Does the provider invest in a building-wide infrastructure, or restrict to a point solution
per channel or somewhere in between? Initially there will be few HD channels in any
playout provider but, given predictions such as, ‘100 HD channels across Europe by
2010’, it will be very interesting in four to five years time to look back at the roll-out.

We tested many cable types, cable lengths and patch panels, to see just how far we
could get HD-SDI around our building. Initially, we were expecting to convert to fibre
in many places, but this turned out not to be necessary. The added benefit was lower
cost and less potential points of failure within a chain.

There are choices to be made regarding the archiving of HD content. Initially, with
low volumes, it may be simpler to keep transmission copies on the playout storage,
or on master tapes and ingest every time it is to be repeated. But to save expense,
as with SD, automatically archiving through a media management system onto data
tape is probably the longer-term higher-volume solution. Apart from bandwidth within
a media management system, the playout/archiving workflow should be very similar
to that of Standard Definition.

One of the bigger issues in HD is monitoring. Can the playout provider see any of the
artefacts anyway? What monitors should be used and do they hide or exaggerate the
problems? Also to consider is if you should monitor every point of a chain in HD or
downconvert wherever possible. Another point is if Playout or Engineering is the
correct place to monitor ‘quality’ and if £30k really needs to be spent on a grade 1
CRT. Is a CRT the most appropriate display device anyway, considering viewers will
probably be watching on LCD or Plasma. A balance must be struck and Red Bee
Media has spent a lot of time debating and testing theories around this.

As HD-Ready is a wide specification standard consideration needs to be given to
how the viewer experience should be mimicked by the playout company. We have
seen both good and not so good displays within the HD-Ready label. The interlace to
progressive and pixel-mapping conversions within the device can expose issues
here.

5.1 audio is another of the major challenges facing HD broadcasters. The HD spec
covers multi-channel sound which, to the viewer, should be similar to that presented
by DVD or games on a surround sound home cinema setup. Issues include the
obvious to the not so obvious. How do you insert a dual mono end-credit voice-over
into a 5.1 stream without collapsing the surround sound? Will it become possible to
insert without returning to individual discreet tracks.

How does playout do a mix between an upconverted SD stereo item and a 5.1 true
HD item, and can metadata actually work within a mix? What will the STB do,
assuming it has been set up correctly in the first place?

Where does multi-languaging come in – embed the language within the 5.1 or send
language tracks separately to the 5.1 M&E tracks around the playout infrastructure.
Do we then combine before distribution or will a STB be able to mix appropriately?
These will be issues for channel owners who broadcast the same content over many
European territories.
For the BBC trial, Red Bee Media and BBC Research and Development chose Dolby
E as the content delivery and internal playout standard and AC3 for the delivery to
the STB. Issues that had to be thought through include encoding/decoding delays
and display processing delays. The STB or home cinema amplifier will include a
delay to compensate for the processing delay of the display.

For those without home cinema, a stereo mix is created by the dolby decoder in the
STB, hence the need to preserve accurate metadata throughout the chain.

To allow for a stereo Audio Description mix to be created prior to encoding, other
options include using tracks 7 and 8 of Dolby E to transport the AD mix.

The last thing to question is whether it is necessary to provide regulators with a
higher bitrate HD Record of Transmission (off-air recordings) including the ability for
surround sound re-creation, or just to downconvert and compress to the equivalent of
a stereo SD channel.

Contribution and Distribution

There are two areas of concern - incoming and outgoing. Incoming live events, and
the bandwidths and number of encoding/decoding points used will be a balance
between cost and quality. These issues have yet to be resolved in the UK, largely
because STB manufacturing delays have meant we can only guess at what ‘quality’
will be like for the entire end-to-end process.

For outgoing connectivity, do we encode a single channel to a fixed bitrate and send
a transport stream to the multiplex or uplink site, or do we send HD-SDI? It will be
interesting to see the cost differences, although from discussions with some
telecommunications companies, this may well not be such an issue.

What space segment bandwidth do you buy, what is the best or most appropriate
bitrate to use, what is the most a set top box can decode before it gets overloaded
and crashes, and what will lead to viewers complaining? Bearing in mind that the
average display size is increasing from the mid 20-inches to the mid 30-inches, and
that viewing distance will not increase to compensate, artefacts will become more
obvious. Standard Definition transmissions on a large domestic display can look poor
– HD in this instance is more about catching up with the quality of Blue-Ray/HD-DVD
and games consoles and reaching the raised consumer expectation.

Will Satellite owners be able to kick-start HD by selling quality rather than bandwidth?
The entire industry will benefit if the highest costs of launching an HD channel can be
reduced, but at the same time keep quality high. A quality experience will drive HD
uptake and speed up channel launches.

What are the quality differences between 20Mbs, 15Mbs, 10Mbs and 5Mbs and
when has or when will each be a reasonable standard, based on encoding
improvements? How much of a difference could the addition of Stat-muxing make
when there are a lower number of channels within a transponder for the multiplexing
system to compare against?

Finally, when can DVB-S2 be used, and the increase in quality for the same bitrate it
generates, and when can it not. This depends on the HD set top boxes, but also
whether your HD channel is multiplexed with SD channels. (The whole multiplex is
modulated S2 not the individual channels, so if there are SD channels within an S2
mux, then those SD channels would not be viewable on an SD STB.)

Will we get to the ‘similar bandwidths to SD channels’ that encoder manufacturers
are aiming for, and if so, does that make SD redundant, or will we then be able to
transmit 1080p.

Other Delivery Platforms

As with SD, content may need to be re-purposed for other delivery platforms. Editing,
re-versioning, transcoding and DRM for Broadband delivery, IPTV, Mobile Phones,
PVR’s plus the EPG element allowing viewers to find and choose content – ie to get
the most value out of a piece of content, are essential. This should be straightforward
for a standards agnostic media management system. The provider must be able to
help the content owner to get the maximum out of their content.


Viewers

Viewers are the most important part of the High Definition chain, after all, they are
who we should really be trying to impress. An interesting question we heard recently
‘is HD appealing just to a minority technically aware audience or will HD appeal to a
wide spectrum of viewers?’ Initially, due to the cost of quality displays and the home
cinema add-on’s, it will almost certainly be affluent early adopters who want to watch
sports, films, natural history or drama in HD. These viewers will still keep their
second and third sets as SD and without surround sound – probably for the kitchen
and kids bedroom. But as costs of launching HD channels reduce and audience
figures rise, a wide variety of channels will evolve and will slowly replace existing SD
channels.

What does HD-Ready mean? We should all know by now, that it is a very good but
reasonably wide specification. However, unfortunately some manufacturers are
marketing ‘Ready for HD’, ‘HD Compliant’ and ‘HD Compatible’. What can be done to
ensure we understand what viewers will be watching on, if they start to complain
about picture and sound quality?


Summary

So, is it time to outsource HD playout?

Certainly channel owners wanting to launch a few HD channels, or anyone wanting
to launch while the technology is still immature, will benefit from the scale of a large
playout operation where time can be spent on the essentials of HD – understanding
the clients requirements and business models, understanding the process whereby
the client acquires content, understanding the technology hurdles and then spending
money training operators and engineers in-depth. A multi-client area, where scale of
infrastructure can be used to lower costs, is essential while revenues from HD are
limited by the lower viewing numbers.

Playout providers must understand the complete end-to-end process of HD –
including acquisition/post production/distribution to various platforms (including IPTV,
broadband, satellite and potentially terrestrial). Where the perceived quality of SD
can be ‘average’ to ‘good’, it is very easy to reduce the quality of HD through
inexperience. Viewers – the most important people within HD - will be buying larger
displays, which will amplify artefacts. We must ensure that they get the experience
they are expecting, having spent considerable sums on large HD displays, set top
boxes and subscriptions. These early adopters will dictate uptake by word of mouth –
we must give them quality, so as to increase audience demand.

HD is not as straightforward as SD has become. All of the questions noted elsewhere
in these white papers, and in the BKSTS seminar, show that while HD can look great,
achieving this is as difficult as working with and around the analogue technologies of
the past.


Written by:
Chris Howe, Chief Technology Officer
Steven Abbott, Technology Manager, HD
TVB Europe White Paper

HD Subtitles
Red Bee Media is working with BBC R&D to help set standards in the languaging
and Access Services areas of High Definition. We have looked at subtitle issues in
our 4 month HD test channels in 2005, and are working with BBC Television and
BBC R&D in the BBC's HD trial service due to launch in a few months. In this paper,
we will state a few of the problem areas and the possible solutions.

Subtitles:

Whilst subtitles are used in the UK mainly for the deaf and hard of hearing, as a
playout provider we must also consider their multi-language use. Standards should
be defined with this in mind. Until recently, 25 frames per second was the standard
for broadcast in Europe – this can now be 25 or 50, so how will preparation systems
adapt? A transcode from an HD program file to a browse format suitable for the
subtitle originators may take longer. With display sizes generally increasing, subtitles
now can look invasive, so can more text or more appropriate positioning be
considered. How will the multiplicities of display devices and set top boxes deal with
subtitle data? In the future, can we present data as a constant stream, therefore
providing the subtitle for each frame, to allow instant availability?

Work initiated by the W3C Timed Text Group for the authoring, conveyance and
archiving of time-critical textual content now needs to be taken through by relevant
bodies to practicable completion. Meanwhile we are using standard SMPTE ancillary
data protocols to encapsulate such subtitle information in the HD-SDI signal and the
DVB Subtitling standard (EN 300 743) has recently been extended to accommodate
the transmission of bit-mapped subtitles at HD resolutions to the home HDTV
receiver.


Audio

With 5.1 audio now becoming a standard for HD, how do broadcasters, audio
description and language houses create and manage these files together with the
relevant metadata?
If we use Dolby E, how do we manage the frame delays of processing within Dolby
encoding/decoding and what happens at the viewers display where there is
additional processing which can put audio out of synch.


Signing

Will there be a future set top box that can hold video or data information relating to an
in-vision signer, and will it be combined onto the channel output? This is future work,
which will need standardisation.


Written by:
Chris Howe, Chief Technology Officer
Steven Abbott, Technology Manager, HD