NG
I think there's some statistical thing that makes it practical to encode as late as possible, taking into account the data on all the other channels on the same, umm, <thing> (mux?).
That's statmuxing isn't it - where you share a pool of data between channels dynamically, varying the bitrate for each based on content (so if one channel is showing something tricky to compress whilst another is showing something easier to compress they give the former more bitrate and the latter less). Traditionally this has required the encoders to be co-sited, though I believe remote encoding with inter-site signalling can also be used now?
And I think for a related reason all the regions are slightly offset from each other in time deliberately. You couldn't guarantee every decoder would get the switch right anyway.
That's the case for the sub-regions I believe (where there is potentially a two-stage coding delay to allow them to opt-out of their "mother-region" rather than network) However you could probably get round this with some uncompressed circuits (as currently used for backhaul I believe)
Additional delays are sometimes introduced to assist statmuxing (or is this bobbins?) when you permanently broadcast network content on multiple channels (as is the case with the BBC approach to regions - which is 24/7 broadcasting of a feed which varies little during the day) It wouldn't be needed in the SVT scenario as the various feeds all carry different content.
Ignoring the nations, which are different, and possibly Jersey (as it has complex backhaul arrangements to Plymouth I believe?), it should be possible to engineer a similar approach - given that there are uncompressed circuits from each regional centre back to London these days. (The distribution of SD BBC One network TO each region is still via an MPEG2 circuit - 9Mbs ISTR - but now the encoding is done in London not on-site, the feed of the "dirty" BBC One - with the regional content added - is uncompressed 270Mbs SDI-rate - so that encoding can take place centrally. So the connectivity is potentially there to adopt a system like SVT I guess. AIUI it is what ITV does in pres terms - though not encoding/uplink)
You may need to use 2 transponders and duplicate the network feeds - as the UK has more regions than Sweden - as this approach only works when switching within a single transport stream (which with most DVB-S/S2 systems means a single transponder - though more esoteric DVB-S2 approaches allow multiple transponders to be stitched together to create a very big transport stream)
I guess this is where it wouldn't scale to HD, and I guess it is also based on it being acceptable to reduce picture quality for regional content?
noggin
Founding member
Quote:
I wonder if SVT and NRK encode centrally and/or can GOP align the encoders and PID switch on a GOP boundary?
I think there's some statistical thing that makes it practical to encode as late as possible, taking into account the data on all the other channels on the same, umm, <thing> (mux?).
That's statmuxing isn't it - where you share a pool of data between channels dynamically, varying the bitrate for each based on content (so if one channel is showing something tricky to compress whilst another is showing something easier to compress they give the former more bitrate and the latter less). Traditionally this has required the encoders to be co-sited, though I believe remote encoding with inter-site signalling can also be used now?
Quote:
And I think for a related reason all the regions are slightly offset from each other in time deliberately. You couldn't guarantee every decoder would get the switch right anyway.
That's the case for the sub-regions I believe (where there is potentially a two-stage coding delay to allow them to opt-out of their "mother-region" rather than network) However you could probably get round this with some uncompressed circuits (as currently used for backhaul I believe)
Additional delays are sometimes introduced to assist statmuxing (or is this bobbins?) when you permanently broadcast network content on multiple channels (as is the case with the BBC approach to regions - which is 24/7 broadcasting of a feed which varies little during the day) It wouldn't be needed in the SVT scenario as the various feeds all carry different content.
Ignoring the nations, which are different, and possibly Jersey (as it has complex backhaul arrangements to Plymouth I believe?), it should be possible to engineer a similar approach - given that there are uncompressed circuits from each regional centre back to London these days. (The distribution of SD BBC One network TO each region is still via an MPEG2 circuit - 9Mbs ISTR - but now the encoding is done in London not on-site, the feed of the "dirty" BBC One - with the regional content added - is uncompressed 270Mbs SDI-rate - so that encoding can take place centrally. So the connectivity is potentially there to adopt a system like SVT I guess. AIUI it is what ITV does in pres terms - though not encoding/uplink)
You may need to use 2 transponders and duplicate the network feeds - as the UK has more regions than Sweden - as this approach only works when switching within a single transport stream (which with most DVB-S/S2 systems means a single transponder - though more esoteric DVB-S2 approaches allow multiple transponders to be stitched together to create a very big transport stream)
I guess this is where it wouldn't scale to HD, and I guess it is also based on it being acceptable to reduce picture quality for regional content?
Last edited by noggin on 30 March 2013 11:11am
OD
When you put it that way, I actually agree. While such pauses and "programming continues in X minutes" may seem clunky in this fast-paced day and age, what's really the problem?
Don't you dare take my relaxation moment away from me! I love the nice red slide with bubbling sounds. A moment of peace and quiet in an otherwise hectic day. Also a lot more entertaining than regional news!
When you put it that way, I actually agree. While such pauses and "programming continues in X minutes" may seem clunky in this fast-paced day and age, what's really the problem?