Just wondering by broadcast needs 50mbs. One of the things that comes to mind is all the re-encoding videos has to do to get to ones TV set from cable, satellite, over the air. So I'm thing that the high bitrate holds the image together for that reason. Am I right?
Broadcasters don't "need" 50mbit/s, it's mainly a baseline (or a dead term if you may), used since the SD era (and then migrated to the HD), and at the end of the day each broadcaster sets his own specifications. But in general you're right, they ensure a certain level of quality (esp. when using long GOP codecs for HD), but depending on the nature of the program (e.g. Deadliest Catch-gopros) these specifications can be overlooked. I actually wrote an essay on this subject (for my Msc), here is a more technical explanation if anyone's interested (page 69,70,71,..) : http://books.google.co.uk/books?id=jc8EZwEuEIoC&pg=PA69&lpg=PA69&dq=50mbit broadcasting quality?&source=bl&ots=2GCbTBtNVE&sig=9qd--06dk_zbPua99DOJ7R7OVtM&hl=el&sa=X&ei=LlxCT8TGL8rD0QXG36CPDw&redir_esc=y#v=onepage&q=50mbit broadcasting quality?&f=falseAlso , as stated by a dvxuser member (Nugat): "In Europe it's fairly simple. EBU (European Broadcasting Union) has the resources and knowhow to do research in the name of its members. In the 2008 R 124R recommendation , based on thourough scientific examination, EBU recommends for production and archiving : --422 sampling --50 mbit for long GOP --100mbit for intra codecs --8 bit is ok, but 10bit for some types of pictures Members (eg BBC) can follow and/or modify the requirements with additional conditions. Non-members (a lot of non-public networks) can do whatever they want. PS EBU also recommends progressive picture (720/50p for broadcast) but here nobody follows. 124R is an interesting read". Practically, HD broadcasters like BBC or Discovery Channel wanted to avoid using footage from cameras with low bit-rate long-GOP codecs as a general rule. On the other hand long-GOP codecs are getting better, and managing HD footage is a pain in the a**, so broadcasters have started asking camera manufacturers for 25 Mbits/s long-GOP codecs for a faster workflow. Confusing times..
Curious on this as well. It probably comes down to quality and the standard that they want to stick to (i.e. the Mb/s equivalent of SD). I have seen GoPro footage on Top Gear that seems to handle perfectly well amongst the higher bitrate cameras. EDIT: for some reason this was posted an hour or so after I wrote it. Only thing to add is that the BBC requires at least 75% of it's programming to come from 50Mbit/s cameras. The rest can be from anything. However, if the content is more important than the quality of the image (panorama hidden cameras for example) then exceptions can obviously be made.
thx @Archie, I didn't know the BBCs "75%" policy, interesting. I'd also like to hear some "real life" examples and any additional info (or corrections).
Not my example but this is the article that I originally found the 75% rule (although after re-reading I'm not sure if it is a proper rule). http://www.definitionmagazine.com/j...rove-video-dslrs-on-a-case-by-case-basis.html It's a great site and has some really good behind the scenes articles that go into details about the kit used which I find that other BTS articles don't in fear of scaring off some readers.
I don't do broadcast, but a couple of friends are ex-BBC engineering. By plying them with beer, I sought to find out if there was a justifiable reason, or whether this was like the 'must be 1/2" or bigger, 3CCD' note for SD, which was more about preventing Production Companies cheaping out on image quality... That the Beeb puts in these barriers on purpose to maintain quality standards. Of course, if you shoot a major documentary on Z1s and edit them in your garage, but end up with 90 mins of stunning television, the Beeb doesn't care one jot. However, I've heard from three sources that the 50 Mbit 4:2:2 is important because of the god-awful route broadcast work takes from delivered final programme to end viewer using Freeview Digital TV reception. Compressing a compressed picture that's been compressed from compressed rushes leads to very dirty pictures. The method used for compression throughout the production chain is (I am standing by to be corrected here) 'Discrete Cosine Transformation' which tends to look for little bits of detail and 'enhance' them, whilst finding areas of bland uninteresting detail and replacing them with a broad brush 'wash'. Now, you take 4:2:0 rushes in 8 bits, and edit them - in the process, grading them by stripping the 4:2:0 8 bit images of their full dynamic range. No visible damage done, and yes, if you get an oscilloscope out, or tweak its electronic nipples off to prove a point, there's some damage done. Now, encode that again for distribution, encode it again for hoofing off to transmitters, encode it again to mux in everything else, then add lots of error protection, chop it into 1080i whether it likes it or not, and then squeeze it through a 2 mbit (i.e. HD Vimeo) hole, you've given the original footage many beatings, and all those visually invisible artefacts are now beginning to show. All those invisible edges in your 8 bit, all those barren areas of tone, are either enhanced (think 'in camera sharpening') or flattened to buggery upon every iteration. Edges get covered in mosquito noise, flat areas end up looking like a picture made of lego. The effect is logarithmic and exponential. It is, and I quote the appropriate engineering term here, 'fugly'. If your broadcast chain is only 3-5 steps, then 35Mbits is fine. The BBC - and other big networks - can't quite manage that (politics, bureaucracy, due process) and therefore need just one step more than is sane in order to maintain the quality they've promised. Of course, the technical niceties are slavishly adhered to, whereas white balance, focus and audio quality seem to be quietly leaking away. Hey, An XF305 does broadcast quality! Putting it in AF/AWB and hand held doesn't make the picture any better, but the producers can check a little box marked 'yes'. And whilst I'm ranting and frothing at the mouth, did anyone clock Jonathan Harrison's presentation at the UK's BVE show? LED lamps are not all made equal, big issues lurk, expensive mistakes have been made. But broadcasters have engaged the amazing and rightly hallowed Alan Roberts to write up a spec for LED lighting. Never mind your camera, dear chap. Is your lighting going to be broadcast spec? Sorry, if you're not using KinoFlo or LitePanels, your name's not down, and you're not coming in.
Funnily enough I was there too! Had to leave just before the end so I could get to an Arri production skills workshop on time. Whilst a bit of it was over my head I thought it was a really interesting point about a topic that I had never really thought about but it makes sense that if a light is not emitting certain frequencies then how can you expect the camera to then see them. If I remember correctly it's all about the R9 and R13 values of the CRI (was it CRI?). On a side note this forum is already reaping rewards for me because it's this sort of knowledge that you just don't pick up online without hours of searching! Thanks for the info Matt.
"However, I've heard from three sources that the 50 Mbit 4:2:2 is important because of the god-awful route broadcast work takes from delivered final programme to end viewer using Freeview Digital TV reception. Compressing a compressed picture that's been compressed from compressed rushes leads to very dirty pictures". You are absolutely right, and in addition, many Freeview's channels are still using 16-QAM modulation (instead of 64), which essentially makes the picture quality almost unbearable (lower QAM modulation has less "symbols" and a lower bit rate, essentially offering less channels and usually worst quality, but on the other hand is more reliable and practically reaches more houses). Here's a modulation table for reference:
IIRC, it began with T, and can't for the life of me remember what came next. CRI and the Mystery Broadcast Quality Txx will probably like f stops and T stops. If you can tell the difference, you may enter the Pro Pub.
Decided to google it this morning! TCSXX (test colour samples) is the format and TCS09 (strong red) and TCS13 (Light pink) are the ones that are lacking with a lot of LED lights. Taken from wikipedia but essentially highlighting what was said at BVE:
