In Part One we briefly looked at the EBU Rec. 118 tiering structure for broadcast cameras.
In this part, we’ll take a look at some popular cameras and see where they fit. Finally, we’ll see if we can adopt some general practices to ensure our camera is of broadcast quality.
Is having the right codec enough?
No. Apart from codecs, EBU also tests cameras systematically for the following:
Generally, the noise level must be above 46 dB for 3CCD cameras and 50dB for Super35-type cameras:
|Tier LS||Better than -50dB @ 0db gain*|
|Tier SP||Better than -42dB @ 0db gain|
|Tier 1||Better than -46dB @ 0db gain|
|Tier 2L||Better than -44dB @ 0db gain|
|Tier 2J||Better than -44dB @ 0db gain|
|Tier 3||Better than -40dB @ 0db gain|
*0dB gain for CMOS cameras is their native ISO.Sensitivity
In short, this test blasts a Kodak gray card with 2000 lux (typical studio lighting is 1000 lux). The camera takes an exposure reading at 0dB (base ISO), and we get an f-number.
This f-number should match what we might get with the same experiment without light but with a white card (89.9% reflectivity).
Exposure range (Dynamic Range)
The more the better. The EBU recognizes that an experienced DP can make low dynamic range look good.
Already given in the chart in Part One.
In their own words:
Cameras should not exhibit high levels of aliasing, at any spatial frequency. Spatial aliasing is evidence that the lens is passing spatial frequencies to the sensor, at frequencies beyond the limits of the camera. This will normally not be acceptable for cameras in Tiers LS, 1 and 2L, although some may be permissible in Tiers SP and 3.
The EBU decides what constitutes ‘high levels of aliasing’ on a case by case basis.
|EBU R 132||100 Mbit/s AVCi or higher||2L and above|
|50 Mbit/s MPEG-2 (inter) or higher|
|35 Mbit/s MPEG-2 (inter) or higher||2J|
|‘Pro-sumer’||35 Mbit/s AVC-HD or higher||3|
|H.264 based codec||4|
As is evident, AVCHD and H.264 codecs are not preferred at all. This is an interesting scenario, since HDCAM SR (SStp) is based on MPEG-4, and XAVC is based on H.264 AVC.
Don’t try to make sense of it. Just go with the flow. The complete methodology and details for testing is available in the Tech 3335 documentation.
Is your Camera Broadcast-worthy?
To find out whether your camera is broadcast-worthy, first check which Tier it belongs to – the prime indicators are the codec, resolution and sensor size.
Here’s a list of cameras, codecs, resolutions and the tier I think they belong to (I could totally be wrong, so don’t use this as a reference):
|DSLRs with 4:2:2||2J|
|DSLRs without 4:2:2||3|
|GoPro Hero 3||SP|
I’ve listed cameras purely based on the internal recording codecs, except for DSLRs. To be honest, just because you use an external recorder to get an 8-bit 4:2:2 from a DSLR doesn’t mean it will be acceptable. See Nikon D800 fails but D4 passes EBU Test.
There are cameras that don’t fit into the EBU scheme of things. The Red Epic is one typical example.
Regarding the Red Epic, you can find their notes in the EBU Tech3335 Supplement 004. Here’s the gist of it:
It is difficult to place this camera within the tiering structure of EBU R.118. If it is used in 5k mode, it qualifies easily for the LS category since the noise level is acceptable at ISO 800 and the sensor size is a little larger than 1”. Arguably, the 4k mode could also qualify for LS.
However, in the lower-resolution modes, the image format size reduces proportionally, thus in 2k mode, the image format is only 10.39×5.84mm, only slightly larger than the conventional 2/3” format of 9.6×5.4mm, and the resolution is little better than 1360×810. Therefore, it properly qualifies for the SP tier, Special Purposes, where reduced resolution and noise are traded for high-speed shooting.
The Blackmagic Cinema Camera is even more confusing. On the one hand, its sensor is smaller than m4/3 (1/2″), but the imagery and codec are exemplary. Furthermore, because it is a 2.5K camera, it easily resolves 1000 lines for 1080p broadcast.
Would it be unacceptable? Or would it be placed along with the GoPros (SP class) as a special case? Who knows? If and when the EBU tests it, we will have a general idea of its viability for broadcast content.
The problem with tiering in general, as I’ve outlined in Part One, is that technology is too fast-paced and broad for this methodology to work. I believe, instead of wasting time and resources testing cameras and technology, let the final content speak for itself.
All that remains is to check the final file format for signal-strength, color and broadcast-worthiness. In this sense, even H.264 video is acceptable, since actual broadcast signals (what you see at home) is far worse than it!
All this is confusing, what do I have to do to guarantee my work is ready for broadcast?
I did warn you that this is an unproductive exercise, didn’t I? Anyway, it would be a waste to stop without any tangible solution. Surely there must be some way to tell if a camera is broadcast-worthy?
Here’s what I believe to be good enough specifications for broadcast quality:
- 1/2″ sensor (3CCD) or m4/3 (CMOS) and above
- 50 Mbps 4:2:2 Interframe codec or 100 Mbps 4:2:2 Intraframe codec
- Full 1000 line resolution (DSLRs usually don’t resolve 1000 lines even at 1080p)
With these specifications, you have a much better chance of not having your project rejected because of your camera. That’s the best that can be done under the circumstances.
If you need more specific help, search the EBU documents to see if they’ve covered your favorite camera.
Very Important: Don’t assume your camera is broadcast-worthy just because the EBU (or anyone else, including me) says so. The broadcast industry is huge, and it is not mandatory for a network or broadcaster to follow these guidelines strictly. Ultimately, with the changing nature of television, and the advent of Internet TV, I believe these guidelines will be less and less relevant.
Have you ever had to worry if your camera is broadcast-worthy? Or, have you ever had a project rejected because your camera wasn’t broadcast quality? If yes, please share your thoughts and experiences.