(2016-04-15, 10:21)wrxtasy Wrote: My Mediainfo inspection proggy, looks like it not giving enough metadata detail to tell.
I assumed it was interlaced, but I now suspect Progessive. Whatever uses less Transmission Bandwidth I would imagine..
Err... No. There is no option to use interlaced for 2160 content - the standards are progressive only.
AIUI the bandwidth benefits for interlacing were mainly in the production chain, where signals are uncompressed (or very lightly compressed). When we started with HD, handling 1080/50i 4:2:2 content required >800Mbs bitrates to be used (higher for live signals as they have H+V blanking). That would have doubled to >1.6Mbs for 1080/50p. In the mid-90s when first gen digital HD standards were being standardised, that difference was huge. This is why we ended up with 720/50p and 1080/50i (and 1080/25p in production) for our dominant HD production standards. This isn't just a 'bandwidth for the signal problem', it's processing issue too. Cameras, Switchers, Routers all needed to cope with twice the data rate. This wasn't feasible. In reality it took many, many years for 1080/50p to become a practical production standard (even when cameras arrived that would output 1080/50p, you had to connect two BNCs to the side of the camera, as what came down the main camera fibre or triax to the control room was still 1080/50i or 720/50p due to bandwidth constraints on the cables that connected the camera head to the camera control unit)
However for transmission the difference between 1080/50i and 1080/50p in bandwidth terms is actually far less than double. The EBU here (figures for HEVC) suggest somewhere between 10% more required for 1080/50p than 720/50p, and with older encoders something like 20% for 1080/50p instead of 1080/50i. (So encoder quality does play an issue - but losing interlace is also a very real aim - as deinterlacing is processor intensive, and nothing much more than guesstimation as you are trying to recreate information that was thrown away. The spectrum folding of vertical and temporal resolution that interlacing introduces is also far from ideal)
So actually it was the production systems that dictated the choice of 1080/50i rather than 1080/50p NOT the encoder efficiency. The reality is that if we'd been able to produce 1080/50p content in the 90s, we'd have started broadcasting 1080/50p almost certainly. Interlacing is really a legacy system designed to solve the problem of bandwidth reduction in the uncompressed and analogue domains.
This may be interesting reading :
https://tech.ebu.ch/docs/techreports/tr036.pdf It's a recent EBU publication looking at how Europe will cope with the loss of the 700MHz TV band (after already losing the 800MHz one), and discusses the efficiency gains of H264/AVC and H265/HEVC over MPEG2, and also the benefits gained by working VBR with a statmuxing system (where bandwidth is dynamically shared between services based on coding needs for delivering similar levels of quality across your channels. Fixed bitrate doesn't deliver fixed quality)
Also - there is some interesting discussion to have about higher frame rates not requiring massively more bandwidth to encode either. You don't need twice the bandwidth to encode 1080/50p as you do for 1080/25p in H264 or H265 (assuming you aren't using Intra frame coding where every frame is coded separately - and that approach is never used for broadcasting to the consumer). Because capturing frames at twice the frame rate captures motion in a more detailed manner (you have twice the data set from which to track motion), it assists the encoders in tracking vectors of the various image elements (also there can be less motion blur on each frame too, which makes block matching more effective). If you can do this, you can improve picture quality or reduce bitrate.