In broadcast, the safe area is the part of the frame where you know everything will be seen, whatever the screen. That’s the idea behind this section of the blog: a safe place to talk about the technical side of audiovisual QC, explained clearly and without taking anything for granted.
It all started (without me knowing it) in the second year of my degree in Audiovisual Communication at the Complutense University of Madrid. We had a subject called Tecnología de los medios audiovisuales (Audiovisual Media Technology), taught by three professors, each one in charge of a different part. At the time it felt tedious; today I remember it fondly. When the codecs and containers part arrived, they handed us a pile of photocopies with an endless list of formats, their features and their uses. Photocopies! Yes, photocopies, with that typical grey tone, specks of dust and the odd stain. We had to study every single one of them, many already obsolete back then (this was some fifteen years ago), with names like D1 or D2 (tape formats rather than codecs, but that’s another story), alongside others like MKV that are still very much alive.
What started as something I just wanted to get over with soon fascinated me, and I ended up making it part of my job. Now I’d like to share what I’ve learned along the way; hopefully in a friendlier format than those photocopies.
Broadcast mode
Whenever I hear the word «broadcast», something switches in my head. I slow down, I open a longer checklist, and I go into what I call (half joking) broadcast mode. It doesn’t mean digital deliveries get less care; content is always checked under the magnifying glass, wherever it’s going. It means there’s one more layer on top, and that layer matters just as much as the content: the form.
When a video is going to end up on social media or a website, QC is mostly about content: is the text right, is the translation in place, does the image look good, does the audio work. We check the format too, of course, but the weight is on what the viewer sees and hears.
Broadcast is different. The content still gets the same attention, but the form becomes a second QC: the specifications, how the file is presented, the timecode, the audio bit depth, the levels, and a long etcetera. A file can look perfect and still be rejected because one of those values is out of place. That’s why a broadcast QC takes longer; it’s two QCs in one.
A real example. You can be checking a file where the content is perfect and every spec is in place, and then you get to the audio. The delivery asks for EBU R 128, so the true peak can’t go above −1 dBTP, and this file measures −0.9 dBTP. It’s a difference nobody could hear, but the file doesn’t meet the spec, so it goes back for a new render and the QC starts again from scratch. And that’s the catch: sometimes, when the new render comes out with the true peak fixed, other values have moved along the way. Not always, but often enough that you can never take anything for granted. That’s why every new render gets checked in full, broadcast or not.
This is the first post of a series about that layer. Here I’ll go through everything briefly, like an index; in the next posts I’ll go deeper into each topic, one at a time.
Why so strict?
Most of these rules were born in analogue television, when the signal was limited and everything had to fit inside it. The video signal, for example, lived between 0 mV (black) and 700 mV (white); anything outside that range could cause problems on air. Old CRT televisions also cut off the edges of the picture (what we call overscan), and interlaced video and frame rates like 23.976 come from those days too.
The digital era removed most of those limitations, but the rules stayed. The EBU itself says it clearly in its R 103 recommendation: readings in millivolts «no longer give relevant information» in digital infrastructures. And yet the limits are still there, translated into digital values. In a way, broadcast QC is also a small lesson in television history.
And those rules are not the same everywhere. Each territory has its own: Europe (EBU), the United Kingdom (DPP), North America (ATSC and the FCC), Australia (Free TV), Japan (ARIB), and so on. On top of that, each channel and each platform has its own delivery specification. And the same content can be delivered in different ways depending on whether it’s SDR or HDR, whether it goes to a TV channel or is published online with broadcast specs.
If there’s one rule that sums up broadcast QC, it’s this one: always read the delivery spec.
What we check in the form
1. Container and codec
A simple way to picture it: the container is a box, and the codec is the way you arrange things inside it. Each container is a box with a different shape; a MOV could be a square box, an MP4 a rectangular one, an MXF a trapezoid (and so on). The way you arrange things inside is mostly independent of the shape of the box, although some arrangements are made for specific boxes. ProRes, for instance, almost always travels in a MOV. And some boxes are relatives: MP4 is directly based on Apple’s QuickTime format, the one behind MOV.
In practice, this is the part of QC where MediaInfo becomes your best friend. Among other things, we check:
- Container and codec (and its profile)
- Bitrate
- Resolution and aspect ratio (HD or UHD)
- Frame rate and scan type: progressive or interlaced, and field order
- Chroma subsampling and bit depth (for example, 4:2:2 at 10 bits)
- Colour space (BT.709, BT.2020)
- Timecode track
About frame rates: 23.976 and 29.97 are still common in the United States, while Europe works in 25 and 50. And interlacing is not dead; Channel 4, in its 2023 delivery spec, still asks for 1080i at 25 frames per second.
2. Video levels
Remember those 0 to 700 mV? In digital, they became code values. In broadcast we usually work in legal range (also called limited or narrow range), where black and white sit at 16 and 235 in 8 bits, or 64 and 940 in 10 bits; full range uses every available value. EBU R 103 allows a small margin around those numbers, but going beyond it is a classic reason for rejection.
3. SDR and HDR
HDR is another way of seeing the same content. It’s not more saturated colours, and it’s not a more vivid picture. When HDR became fashionable, we saw a lot of over-processed images, almost expressionist, that were not pleasant to look at; that’s not what good HDR is about.
Good HDR is subtle, and most of the time you don’t notice it. You notice it in specific scenes: a dark street at night where you can still see the details in the shadows, and suddenly a street lamp or a car’s headlights shine really bright. That contrast exists in SDR too, but the lamp is not as bright and the dark areas are more crushed. In my opinion, good HDR is the one you don’t notice; the natural one. (It’s the same in video games.)
On the technical side, the reference is ITU-R BT.2100, which defines two ways of encoding HDR: PQ and HLG. From there come the formats we usually deal with: HDR10 (PQ with static metadata), Dolby Vision (with dynamic metadata, scene by scene) and HLG, very common in live broadcast.
4. Audio
Loudness. There has to be a standard so that different programmes sound at a similar volume, and we don’t have to reach for the remote every time something new starts. For years, commercials competed to sound louder than everything else; it was known as the loudness war, and standards were created to end it. Today each territory has its own target:
- Europe, EBU R 128: −23 LUFS
- United States, ATSC A/85 (mandatory by law since 2012, with the CALM Act): −24 LKFS
- Australia, OP-59: −24 LKFS
(LUFS and LKFS are the same measurement with a different name.)
True peak. Besides the average loudness, there’s a limit for the highest peaks. And here the differences show again, even inside Europe: EBU R 128 allows up to −1 dBTP, while Channel 4 in the UK asks for −3 dBTP.
Track configuration. One file can carry several mixes: stereo (2.0), 5.1, Dolby Atmos, and several languages. It can also carry an M&E (music and effects) track, the base used for dubbing. As a mentor taught me years ago, M&E tracks are not measured against loudness targets, because they are working material, not an on-air mix. About Atmos: soundbars and good headphones have made it more accessible, although to really enjoy it you still need the right setup.
Technical format. The usual broadcast audio is PCM at 48 kHz and 24 bits.
5. Heads and tails
Many channels and platforms ask for a specific structure before and after the programme; others don’t ask for any. Streaming platforms such as Netflix, Disney+, Amazon Prime Video, HBO Max or Movistar Plus+ also have their own specifications. The most common elements are:
- Bars and tone: colour bars plus a reference tone (1 kHz), used to check the video and audio chain. The tone level itself depends on the territory: −18 dBFS in Europe (EBU R 68) and −20 dBFS in the US (SMPTE RP 155).
- Slate or clock: a card with the programme information (title, episode, duration, and so on), often with a countdown.
- Black and silence: a few seconds between the leader and the programme, and at the end.
- Start timecode: the programme has to start at a specific timecode; in the UK, for example, it’s 10:00:00:00.
6. Safe areas
Safe areas were born because CRT televisions cut off the edges of the picture. SMPTE redefined them in 2009 for flat screens: 93% of the width and height for action safe, and 90% for title safe. Today, most TVs cut much less than before, and on YouTube nothing gets cut at all.
So, are they still necessary? I think so. Partly because many flat TVs still apply a small overscan by default (as NAB’s summary of the standard points out), and partly for aesthetic reasons: text pushed against the edges breaks the balance of the image. In my view, respecting safe areas is simply good practice, whatever the screen.
7. Territory requirements
Finally, each territory, country or region has its own regulations. One of the best known is the photosensitive epilepsy test (PSE): in the UK, following Ofcom guidance, programmes must pass it, and a failed test means a rejected file. Other countries ask for specific legal notices or disclaimers that don’t exist elsewhere.
Each of these sections will get its own post in the series, with more detail than fits here.
In a nutshell: broadcast QC is two QCs in one: the content, which always gets the same care, and the form, a set of rules inherited from analogue television that still decide whether a file goes on air or comes back. It takes longer, it asks for more patience, and it has a lot of small numbers to remember. But when it’s done well, nobody notices; like good HDR, like good localization, like good QC.
Further reading
Loudness
- EBU R 128: Loudness normalisation and permitted maximum level of audio signals (EBU)
- OP-59: Measurement and management of loudness for TV broadcasting (Free TV Australia)
- Loud commercials and the CALM Act (FCC)
- ATSC A/85:2026-07, Annex M: loudness and true peak quick reference (ATSC)
- Reference levels in digital audio systems (Sound On Sound)
Video levels and HDR
- EBU R 103: Video signal tolerance in digital television systems (EBU)
- ITU-R BT.2100: Image parameter values for high dynamic range television (ITU)
Safe areas
Delivery specs and territory requirements
- Technical specification for the delivery of programmes (DPP) (Channel 4, 2023)
- Guidance note on flashing images and regular patterns (Ofcom, legacy ITC document)
Containers and tools
- SMPTE RDD 44: Mapping and application of Apple ProRes in MXF (SMPTE)
- MediaInfo (MediaArea)